Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Cari Covell

Thesis of 60 ETCS credits Master of Science in Energy Engineering - Iceland School of Energy

January 2016

Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Cari Covell

Thesis of 60 ECTS credits submitted to the School of Science and Engineering at Reykjavík University in partial fulﬁllment of the requirements for the degree of Master of Science in Energy Engineering - Iceland School of Energy

January 2016

Supervisors: Dr. María Sigríður Guðjónsdóttir, Supervisor Adjunct Professor, Reykjavík University

Mr. Sverrir Þórhallsson, Co-Supervisor Drilling Engineer, Iceland GeoSurvey (ISOR)

Examiner:

Dr. Ágúst Valfells, Examiner Department Head, Mechanical & Electrical Engineering, School of Science and Engineering, Reykjavík University Copyright Cari Covell January 2016 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Cari Covell

60 ECTS thesis submitted to the School of Science and Engineering at Reykjavík University in partial fulﬁllment of the requirements for the degree of Master of Science in Energy Engineering - Iceland School of Energy.

January 2016

Student:

Cari Covell

Supervisors:

Dr. María Sigríður Guðjónsdóttir

Mr. Sverrir Þórhallsson

Examiner:

Dr. Ágúst Valfells The undersigned hereby grants permission to the Reykjavík University Li- brary to reproduce single copies of this project report entitled Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use and to lend or sell such copies for private, scholarly or scientiﬁc research purposes only.

The author reserves all other publication and other rights in association with the copyright in the project report, and except as herein before provided, nei- ther the project report nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author’s prior written permission.

Date

Cari Covell Master of Science Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Cari Covell

January 2016

Abstract Direct use of hot water through renewable energy resources is globally in demand. Thermal energy stored in fractures and pores within geothermal reservoirs contains natural fluids. At times, extracting natural fluids, or hot water in low-temperature areas, can be a challenge. Hydraulic stimulation is one technique to overcome this challenge. Research about hydraulic stimulation methods was done based on theory, fluid treatment, and well testing; in order to see unique trends for low-temperature geothermal applications. Furthermore, a literature review of all hydraulic stimulation applications was conducted to understand reasons for success or failure. In order to predict the effects of hydraulic stimulation before an actual operation, a case study was performed on well HF-1 in Hoffell, Iceland. First, a preliminary production flow model was performed using updated data at the completion of testing in 2014. After evaluating the need for stimulation, a fracture model using MFrac was done in two scenarios with an open-hole packer; injection below the packer and injection above the packer. The packer was placed in a conservative interval of 1070-1110 m depth to isolate the main fracture at 1093 m depth. Injection below the packer failed, therefore results from injection above the packer were only suitable moving forward. Subsequently, MProd software was used to find an improvement ratio after simulating stimulation above the packer. The improvement ratio of 1.096 was then applied to the original production data of well HF-1 and a LPM was performed yet again. Reservoir properties of S, T, II, and PI were calculated and compared to original production data. Results indicated the lumpfit model to be very optimistic and improvement of only 4 l/s flow over a 10 year well lifetime was observed. Therefore, the well is not a good candidate for stimulation. However, improvement was seen which proves the potential for this methodology to be implemented in other low-temperature geothermal areas.

Keywords: hydraulic stimulation, packer, direct use, low-temperature, production. Örvun Borholna á Lághitasvæðum Fyrir Beina Nýtingu Jarðhita

Cari Covell

Janúar 2016

Útdráttur Þörf er á aukinni beinni nýtingu heits vatns frá endurnýjanlegum orkugjöfum. Varmaorka finnst í vökva sem finna má í sprungum og holum í jarðhitaker- fum og getur verið mikil áskorun að nýta þennan vökva úr lághita jarðhitak- erfum. Örvun borholna með vökva og ádælingu (e. hydraulic stimulation) er ein leið til að auðvelda þessa nýtingu. Rannsóknir um örvunaraðferðir, vökvameðferð og borholuprófanir voru teknar saman til þess að fá yfirlit yfir örvunaraðgerðir fyrir lághitaborholur. Ennfremur var tekið saman efni úr heimildum fyrir þær örvunaraðgerðir með vökva sem gerðar hafa verið til að fá yfirlit yfir reynslu af þeim. Til þess að spá fyrir um áhrif örvunaraðgerða með vökva eftir borun var rannsókn gerð á holu HF-1 á svæði Hoffells á Íslandi. Fyrst var einfalt rennslislíkan framkvæmt út frá mæligögnum frá 2014. Eftir að þörfin á örvun var metin, var sprunguhermunarlíkanið MFrac notað fyrir tvö tilvik með pakkara (e.packer) fyrir opna holu, annars vegar þar sem svæðið ofan við pakkarann var örvað og hins vegar svæðið neðan við pakkarann. Pakkarinn var staðsettur á 1070-1100 m dýpi til að einangra aðalæðina sem er á 1093 m dýpi. Ádæling neðan við pakkarann gaf ekki góða raun og því voru niðurstöður fyrir ádælingu ofan við pakkarann eingöngu nýttar í framhaldinu. MProd forritið var notað til að reikna út breytingu á rennsli eftir örvun fyrir ofan pakkarann. Hlutfall rennslis fyrir og eftir örvun var 1,096 fyrir holuna HF-1 og var forðafræðilíkan (LPM) notað til að spá fyrir um framleiðslugetu örvaðrar holu. Forðafræðieiginleikar S, T, II og PI voru reiknaðir og bornir saman við upphaflegu gildin. Niðurstöður úr líkan- inu sýna framleiðsluaukningu um einungis 4 l/s yfir 10 ára líftíma holunnar. Því er ályktað að holan henti ekki vel til örvunar. Sú aðferðafræði sem þróuð var við gerð þessa verkefnis er nýtt framlag til rannsókna á lághitasvæðum.

Lykilorð: Örvun jarðhitaholna, pakkari, bein nýting, lághitasvæði, fram- leiðsla v

Acknowledgements

I would like to thank my supervisor María Guðjónsdóttir for her moral support, encour- agement, and guidance in this thesis; and for being a great mentor throughout the duration of my masters studies. I would also like to thank my co-supervisor Sverrir Þórhallsson for providing useful knowledge and assistance with the MFrac suite; as well as my examiner Ágúst Valfells for reviewing the thesis.

I would like to acknowledge the following Iceland GeoSurvey (ISOR) staff: Guðni Axels- son for guidance regarding the initial stages of ﬁnding a case study; Sigurður Kristinsson for providing references to Hoffell drilling reports; and Helga Tulinius for her help with Lumpﬁt beta.

Furthermore, I would like to acknowledge Magnús Ólafsson from RARIK (Iceland State Electricity) for allowing the use of production data for the case study; and Baker Hughes Inc. for allowing academic use of MFrac suite software.

Additionally, I would like to recognize the Geothermal Resources Council (GRC) Grad- uate Scholarship and the Iceland School of Energy Sustainable Future Scholarship for funding throughout my masters studies.

I express my sincerest gratitude to Halla Hrund Logadóttir, Director of the Iceland School of Energy, for being an incredible mentor and a truly inspiring person. Her unconditional support related to my academic, professional, and personal growth is greatly valued.

I am also very grateful to The GREEN Program as they initiated my passion for a ca- reer in geothermal energy on an international scale, and have continuously motivated me throughout my masters studies.

Finally, I would like to thank my friends Jenn and Angelica for their support from home; my classmates whom have become friends through spending quality time at Miklabraut 64 and Ú214; and my family for allowing me to pursue all of my crazy ambitions. vi vii

Contents

List of Figures xi

List of Tables xiv

List of Abbreviations xvii

1 Introduction1 1.1 Objectives...... 2 1.2 Structure of the thesis...... 3

2 Background5 2.1 Well stimulation theory...... 5 2.1.1 Hydraulic stimulation...... 6 2.1.1.1 Mechanics...... 6 2.1.1.2 Frac ﬂuid treatment...... 9 2.1.1.3 Hydraulic well testing...... 12 2.1.2 Thermal stimulation...... 13 2.1.3 Chemical stimulation...... 14 2.2 Types of hydraulic stimulation...... 15 2.2.1 Air-lift pumping...... 15 2.2.2 Open-hole with a packer...... 17 2.2.2.1 Before deploying the packer...... 18 2.2.2.2 Installing the packer...... 19 2.2.2.3 Setting the packer...... 19 2.2.2.4 Opening the bottom plug...... 20 2.2.2.5 Stimulation...... 20 2.2.2.6 Releasing the packer...... 20 2.2.3 Zonal isolation...... 20 2.2.3.1 Use of a liner...... 21 viii

2.2.3.2 Use of a liner with inﬂatable or swellable packers... 22 2.2.3.3 Plug and go...... 22 2.2.3.4 Multilateral wellbores and sidetracks...... 23 2.3 Environmental impacts and seismicity...... 23

3 Literature review 25 3.1 Hydraulic stimulation applications...... 25 3.1.1 Oil and gas industry...... 26 3.1.2 Low temperature geothermal areas...... 26 3.1.2.1 Reykir hydrothermal system...... 27 3.1.2.2 Seltjarnarnes well SN-12...... 30 3.1.2.3 Other low-temperature ﬁelds in Iceland...... 33 3.1.3 High temperature geothermal areas...... 33 3.1.3.1 Baca, New Mexico (USA)...... 34 3.1.3.2 Latera, Italy...... 36 3.1.3.3 Salak, Indonesia...... 37 3.1.3.4 Mt. Apo, Philippines...... 38 3.1.4 Enhanced geothermal systems (EGS)...... 39 3.2 History of stimulation fracture modeling...... 41

4 Methods 43 4.1 Case study: Hoffell well HF-1...... 44 4.2 Lumpﬁt parameter model (LPM)...... 46 4.2.1 LPM solution methodology...... 47 4.2.2 Initial production modeling of Hoffell HF-1...... 48 4.3 MFrac model...... 52 4.3.1 MFrac governing equations...... 53 4.3.1.1 Mass conservation...... 53 4.3.1.2 Mass continuity...... 54 4.3.1.3 Momentum conservation...... 54 4.3.1.4 Width-opening pressure elasticity condition...... 55 4.3.1.5 Fracture propagation criteria...... 55 4.3.2 MFrac solution methodology...... 55 4.3.3 Stimulation set-up...... 56 4.3.4 Governing model parameters...... 56 4.3.5 Wellbore hydraulics...... 58 4.3.6 Rock properties...... 60 4.3.7 Zones data...... 61 ix

4.3.8 Treatment schedule...... 62 4.3.9 Fluid loss...... 63 4.3.10 Proppant criteria...... 63 4.3.11 Heat transfer...... 64 4.4 MProd model...... 65 4.4.1 MProd governing equations...... 65 4.4.1.1 Dimensionless parameters...... 66 4.4.1.2 Pseudopressure...... 67 4.4.1.3 Trilinear solution...... 68 4.4.1.4 Pseudosteady-state pressure and resistivity solutions.. 68 4.4.1.5 Wellbore choked skin effect...... 69 4.4.1.6 Pseudo-radial ﬂow solution...... 69 4.4.1.7 Productivity increase...... 70 4.4.1.8 Desuperposition...... 70 4.4.2 Stimulation set-up...... 71 4.4.3 Formation data...... 71 4.4.4 Single case fracture characteristics...... 72 4.4.5 Well data...... 73

5 Results 75 5.1 MFrac...... 75 5.1.1 Fracture propagation solution...... 75 5.1.2 Proppant design summary...... 78 5.1.3 Total ﬂuid loss and leakoff rate output...... 79 5.1.4 Heat transfer solution...... 81 5.2 MProd...... 82 5.3 Lumpﬁt parameter model...... 84

6 Summary 89 6.1 Discussion...... 89 6.2 Conclusions...... 91 6.3 Future work...... 92 6.4 Recommendations...... 93

A Open hole packer 109

B Thermal Stimulation in Iceland 111

C EGS Applications 121 x

D Fluid and proppant type properties 135

E MFrac report 137

F MProd report 139 xi

List of Figures

1 Fracture propagation as a result of Hydraulic Proppant Fracturing [14].. 11 2 Fracture propagation as a result of Water Fracturing [14]...... 11 3 Schematic illustration of the setup for air-lift pumping [19]...... 16 4 Diagram for design of air-lift pumping, based on water well experience [26] 17 5 Schematic picture of injection via a packer [22]...... 18

6 Results of production testing of well SN-12, where symbols show observed data one hour into each step and lines show calculated output characteristics [41]...... 32 7 Stimulation methods applied to EGS projects worldwide as of 2013 [56].. 40 8 Rock type and well depth of EGS projects worldwide as of 2013 [56]... 40

9 Locations of geothermal areas in Iceland based on reservoir temperature and geology [19]...... 44 10 Map showing the location of the Hoffell case study area [63]...... 45 11 Location of well HF-1 and some exploration wells [64]...... 46 12 A general lumped parameter model used to simulate water level or pressure changes in a geothermal system. The three tank scenario is shown here [70]...... 47 13 Monitored and calculated water level of Well HF-1 from April 9, 2013 to September 8, 2013 of the long-term production test. Calculated values are those of the LPM, where the left shows the two-tank closed model and the right shows the two-tank open model. Time t = 0 corresponds to April 9, 2013 [61]...... 49 14 Monitored and calculated water level of Well HF-1 from May 9, 2013 to May 8, 2014 of the long-term production test. Calculated values are those of the LPM, where the left shows the two-tank closed model and the right shows the two-tank open model. Time t = 0 corresponds to May 9, 2013.. 50 xii

15 Long-term production test for a one year period of well HF-1. Time t = 0 corresponds to May 9, 2013...... 50 16 Predicted water levels in well HF-1 for the next 10 years for different production rates using the ﬁve month period long-term production test data. Conservative predictions using two-tank closed model are on the left. Optimistic predictions using two-tank open model are on the right [61]. 51 17 Predicted water levels in well HF-1 for the next 10 years for different production rates using the year-long period of long-term production test data. The optimistic two-tank open model is shown...... 51 18 MFrac Pipe Friction Empirical Correlations [72]...... 57 19 Wellbore cross section for Hoffell well HF-1...... 60 20 Velocity of cuttings in mm/s [58]...... 62

21 Upper and lower fracture zone height when stimulated below the packer.. 76 22 Upper and lower fracture zone height when stimulated above the packer.. 76 23 Frac width as a function of frac length for stimulation below the packer.. 77 24 Frac width as a function of frac length for stimulation above the packer.. 77 25 Net pressure measured after stimulation below and above the packer.... 78 26 Frac ﬂuid efﬁciency after stimulation below and above the packer..... 81 27 Leakoff rate after stimulation below and above the packer...... 81 28 Temperature as a function of fracture length after stimulation below and above the packer...... 82 29 Flow rate of Hoffell well HF-1 before and after stimulation...... 84 30 Measured and calculated water level of a long-term production test after stimulation of Hoffell well HF-1...... 85 31 Long-term production test before and after stimulation for a one year period of well HF-1...... 85 32 Predicted water levels in well HF-1 for the next 10 years for different production rates using the simulated year-long period of production test data after stimulation. The optimistic two-tank approach is shown..... 87

33 Cross section of Baker Hughes packer 10.375" OD ISP for formation test [19]...... 109

34 Pressure, temperature, and injection rates during step-rate test of well KJ- 38 [18]...... 114 35 Temperature logs throughout stimulation of well HE-8, where the Novem- ber logs were performed after a 3 month cooling period following drill completion [92]...... 116 xiii

36 Calculated injectivity index for each cycle of stimulation in well HG-1 as a function of volume of injected ﬂuid into the well [18]...... 118

37 Fluid type parameters for WG-11 20 lb/Mgal low viscous gel [72].... 135 38 Proppant type parameters for 20/40 mesh Jordan sand [72]...... 136 xiv xv

List of Tables

1 Improvement ratios in MG-27 [43]...... 29 2 Improvement ratios in MG-35 (based on [43])...... 30 3 Well Baca 23 hydraulic stimulation treatment schedule [49]...... 35

4 Predicted water levels in well HF-1 after 10 years production [m]..... 52 5 Fanning friction factors...... 57 6 Casing Dimensions for Hoffell well HF-1...... 59 7 Hoffell well HF-1 deviation...... 59 8 Rock properties of Hoffell well HF-1...... 61 9 Treatment schedule for Hoffell well HF-1...... 62 10 Pressure dependent ﬂuid loss for well HF-1...... 63 11 Proppant criteria for well HF-1...... 64 12 Heat transfer properties for Hoffell geothermal ﬁeld...... 65 13 Formation data for well HF-1...... 71 14 Well data for Hoffell well HF-1...... 73

15 Fracture propagation solution. Calculated values are at the end of treatment. 78 16 Proppant design summary for stimulation below and above the packer... 79 17 Summary of ﬂuid loss below and above the packer. Calculated values are at the end of treatment...... 80 18 MProd single case fracture characteristics dialog box...... 83 19 MProd production solution for Hoffell well HF-1...... 84 20 Improvement ratios for well test data after stimulation above the packer.. 86 21 Predicted water levels after stimulation in well HF-1 after 10 years production based on year-long production data...... 87 xvi xvii

List of Abbreviations

BMU The Federal Ministry for the Environment (Bundesmin für Umwelt, Germany) BHTP Bottom Hole Treating Pressure CA California DOE Department of Energy ECP External Casing Packer EGS Enhanced Geothermal System ENEL National Entity for Electricity (Ente nazionale per l’energia elettrica, Italy) EOJ End of Job EPDM Ethylene Propylene Diene Monomer FLA Fluid Loss Additive GDY Geodynamics Ltd. GRWSP Geothermal Reservoir and Well Stimulation Program HDR Hot Dry Rock HEGF High Energy Gas Fracturing HF Hybrid Fracturing HPF Hydraulic Proppant Fracturing HWR Hot Wet Rock ID Inside Diameter IDH Idaho ISOR Iceland GeoSurvey LPM Lumped Parameter Modeling MD Measured Depth NEDO New Energy and Industrial Technology Development Organization xviii

NM New Mexico NV Nevada OD Outside Diameter OFEN Federal Ofﬁce of Energy (Switzerland) OR Oregon PRB Permeable Reactive Barriers RSF Reactant Sand Fracking TVD Total Vertical Depth USA United States of America WF Water Fracturing WHP Wellhead Pressure

Nomenclature

A Leakoff area (one face of the fracture) C Total leakoff coefﬁcient

C1 Dimensionless inverse fracture diffusivity

CD Dimensionless wellbore storage coefﬁcient

CDf Dimensionless fracture storage coefﬁcient

Cf D Dimensionless fracture conductivity

Ct Total reservoir compressibility

ctf Fracture compressibility E Young’s modulus F Reservoir aspect ratio G(θ) Fluid loss function H Fracture half-height

Hp Pay zone height

Hw Total wellbore height h Pay zone height J Productivity index k Permeability xix k’ Consistency index kf wf Fracture conductivity kf Propped fracture permeability kh Horizontal reservoir permeability kl Fracture damaged zone permeability kv Vertical reservoir permeability L Fracture half-length lh length of horizontal lateral n’ Flow behavior index NaCl Sodium Chloride

Na2CO3 Sodium Carbonate P Pressure

PwD Dimensionless wellbore pressure

Pwf Bottomhole flowing pressure pi Initial reservoir pressure q Injection flow rate qD Dimensionless flow rate rw Wellbore radius rwa Apparent wellbore radius

Rw Dimensionless apparent wellbore radius S Wellbore skin factor

Sch Chocked fracture skin

Sp Spurt loss coefﬁcient

Sf Fracture skin factor s Laplace space variable T Reservoir temperature t Time tD Dimensionless Nolte time tDA Dimensionless time based on drainage area tDf Dimensionless time based on fracture length xx

tDw Dimensionless time based on wellbore radius

Vf Fracture volume

Vl Fluid loss volume (no spurt loss)

Vsp Volume loss by spurt W Fracture width W(0,t) Average wellbore fracture width W(ζ,t) Fracture width as a function of position

wf Propped fracture width x Lateral coordinate along the fracture length

xf Propped fracture half-length y Coordinate perpendicular to the frac face

ys Damaged zone adjacent to fracture z Vertical coordinate

Greek

αa Leakoff area parameter

αc Leakoff parameter during pumping

αL Length propagation parameter

αp Pressure parameter

αw Width propagation parameter

γf ,Γf Friction coefﬁcients

γw,Γw Width profile coefficients ∆P Net fracturing pressure η Fracture efficiency ζ Dimensionless lateral coordinate θ Dimensionless time µ Equivalent reservoir viscosity τ Time of fracture leakoff area creation Φ Fluid loss parameter φ Equivalent reservoir porosity σ Minimum horizontal stress 1

Chapter 1

Introduction

Direct use of geothermal energy is the oldest and most common form of geothermal utilization [1]. Traditionally, direct use of geothermal energy has been small scale applications by individuals, but more recent developments involve large scale projects in commercial industry. Direct application of geothermal energy involves a wide variety of end uses, such as space heating and cooling, greenhouses, ﬁsh farming, and health spas. Flex- ibility in direct application by use of geothermal energy makes a more attractive option over other means of resource exploitation; such as coal, oil, gas, or electricity.

Geothermal energy consists of thermal energy stored in the earth’s crust [1]. Thermal energy in the earth is distributed between constituent host rock and natural fluids that are contained in fractures and pores at temperatures above some specified reference temperature [2]. In direct use, natural fluid is usually associated with hot water. Sometimes however, extraction of hot water presents several challenges due to possible obstructions in fractures or poor fracture connectivity to the reservoir.

Hydraulic stimulation is one technique to overcome challenges of fluid extraction, which is similar to the more well-known term of hydraulic fracturing. Hydraulic stimulation is the process of injecting fluid into a rock mass at or below the fracture opening pressure, and seeks to induce shear deformation on naturally oriented fractures to increase permeability within the rock mass [3]. Hydraulic fracturing is the process of injecting fluid into a rock mass at a rate and pressure sufficient to form and propagate new fractures [3]. For the purposes of this thesis, the two terms are used interchangeably because applications presented later in this thesis, specific to Chapters 2 and 3, possess qualities of both definitions. However, hydraulic stimulation is the more politically correct term to entitle this thesis, as most theories and applications will be derived from this mechanism. For clarity moving forward, the definition of hydraulic stimulation is used but not for the purpose of 2 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use increasing permeability; rather it is for the purpose of propagating new fractures. This is mostly applicable for the case study performed later in this thesis throughout Chapters 4 and 5.

Hydraulic stimulation can be performed in several types of geothermal areas, including low-temperature, high-temperature, and enhanced geothermal systems (EGS). Low- temperature geothermal areas are defined as having reservoir temperatures below 150°C, while high-temperature geothermal areas are defined as having reservoir temperatures above 200°C [4]. For the purposes of this thesis, EGS areas are defined as having reservoir temperatures less than 200°C, but with very low permeability in the rock mass [5].

Low-temperature geothermal fields, by nature, compose of direct use resources almost entirely for purposes of hot water production; contrary to high-temperature and EGS fields that are capable of co-producing steam for electricity. Therefore, hydraulic stimulation in low-temperature geothermal areas for direct use is a topic of interest, as there is more potential to access natural fluids for a better way to create an almost exclusive network to meet local demand. The knowledge gap lies in predicting the amount of production potential from natural fluids after a hydraulic stimulation operation.

1.1 Objectives

The thesis will focus on geothermal resource extraction for direct use by means of hydraulic stimulation in low temperature geothermal areas. One objective is to research all types of hydraulic stimulation methods available in order to see if a particular guideline is necessary for low-temperature geothermal applications. These methods include air-lift pumping, open-hole packers, and zonal isolation. Furthermore, hydraulic stimulation can be performed using different types of fracturing ﬂuids ("frac ﬂuids") to aid in stimulation; including hydraulic proppants, water fracs, and hybrid fracs.

The next objective is to review the literature available for each type of geothermal area in order to understand reasons for success (or failure).

Another objective is to evaluate the methods most suitable for a case study in Hoffell, Iceland, in order to predict the effects of hydraulic stimulation before the actual operation. This is done by conducting a fracture model using MFrac and a subsequent production model using a combination of MProd and Lumpﬁt beta. Through modeling an optimal hydraulic stimulation, an evaluation of productivity improvement is performed to compare to the literature review. Cari Covell 3

The last objective is to assess a potential need for technological developments within the chosen methods. This includes analyses of improving modeling methodology in order to recommend procedures applicable to low-temperature geothermal environments.

1.2 Structure of the thesis

Chapter 2 discusses background information about three major types of stimulation: hydraulic, thermal, and chemical. Furthermore, hydraulic stimulation is discussed in detail as this is the focus topic of the thesis. Theory of hydraulic stimulation includes several mechanisms of mechanics, frac fluid treatment, and well testing. Methods of hydraulic stimulation are described in three categories: air-lift pumping, open-hole stimulation with a packer, and zonal isolation. Each method includes a guide for conducting stimulation; some of which have multiple options to consider. Lastly, environmental impacts and seismicity are addressed in this chapter. Environmental challenges include the use of chemicals in the fracking fluid, drilling noise, damage to flora, and changes in thermal manifestations. Induced seismicity from stimulation practices has been the topic of many studies throughout Australia and Europe.

Chapter 3 is the literature review for hydraulic stimulation applications. Hydraulic stimulation practices originate from the oil and gas industry, where technologies other than those described in Chapter 2 are discussed regarding potential use in the geothermal industry. Low temperature geothermal areas that have experienced hydraulic stimulation only include sites in Iceland, where emphasis is on the Reykir hydrothermal field and Seltjarnarnes well SN-12. The first experiments of hydraulic stimulation in high temperature geothermal areas were in Japan, but open source information is limited on stimulation programs conducted. Earliest forms of packer and proppant technology in high temperature geothermal areas were initiated by the United States Department of Energy (DOE) in the early 1980s where the Baca site of New Mexico was the first stimulation practice of its kind. Other sites included in the literature review are in Italy, Indonesia, and the Philippines. Enhanced geothermal systems (EGS) are the most common fields for hydraulic stimulation research and application. All EGS sites that have experienced hydraulic stimulation are reviewed in this chapter of the thesis. In addition to performing a literature review of hydraulic stimulation applications, a section is dedicated to the review of fracture simulation modeling through the use of different types of software.

Chapter 4 describes the methods used for various hydraulic stimulation analyses for a case study in the Hoffell low temperature geothermal field of Iceland. Well HF-1 was used to 4 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use target a fracture of 1093 m depth, via two scenarios of hydraulic stimulation below and above a packer. Prior to modeling stimulation, a lumpfit parameter model (LPM) for well production analysis was used to compare more detailed data obtained over a year- long period to a previous study done over five months of data. After obtaining a basis motivation for stimulation in the well, MFrac software was used to model the effects of fracture design and treatment analysis. The well was built in the software based on casing and deviation data, rock properties, and zones of stimulation. Additionally, the treatment schedule, fluid loss, proppant criteria, and heat transfer properties were defined. The MProd software was then used to model production flow after hydraulic stimulation. Formation data, single phase fracture characteristics, and well data were all defined as parameters before simulation. Pressure boundary conditions were also defined from the MFrac output to compare to original production test data of Hoffell well HF-1.

Chapter 5 indicates results after hydraulic stimulation was modeled in MFrac, MProd, and Lumpfit beta software. In MFrac, solutions are included for fracture propagation, proppant design, total fluid loss due to leakoff rate, and heat transfer effects. In MProd, flow rate was analyzed before and after stimulation in order to determine an improvement ratio for the simulated production test. In Lumpfit beta, the improvement ratio was applied to the original year-long production test data of Hoffell well HF-1 in order to determine the success, or failure, of the stimulation. Production over a 10 year lifetime was also included in the analyses of this thesis.

Chapter 6 provides a summary as a means of guided discussion, as well as important conclusions, means of future work, and author recommendations. The summary describes the basis for conducting background and literature reviews of hydraulic stimulation, as well as findings within MFrac, MProd, and Lumpfit beta unique to the case study. Con- clusions about the margin of production improvement with regards to the sensitivity of each software is also provided in this chapter. Future work includes the use of other tools within MFrac and MProd, as well as additional software within the MFrac suite. Finally, recommendations from the author include studies within Iceland and other geothermal fields around the world. 5

Chapter 2

Background

The stimulation of geothermal reservoirs involves the opening up of existing fractures by injecting fluid into a rock mass at optimal high pressures, traditionally performed using hydraulic, thermal, or chemical procedures. Most geothermal stimulations occur as part of well drilling completion programs primarily for three reasons: 1) to get more water flow out of the well for improved economics, 2) clean out drill cuttings that may have caused blockage, as well as or alternatively to 3) increase permeability for connecting fractures between wellbores together to the main reservoir. Depending on the objective of a particular stimulation operation, different hydraulic stimulation methods are available. Each method has a certain form of design and implementation in the geothermal field. With stimulation comes environmental impacts, which are usually associated with negative connotation. However research has shown that several measures can be taken to mitigate environmental impacts related to reservoir exploitation. Seismic history is also discussed in this chapter as microseismic events in Australia and Europe have been the subject of many recent studies, where seismic events are analyzed to understand the underlying mechanisms influenced by stimulation.

2.1 Well stimulation theory

Generally, there are three types of geothermal well stimulation. Hydraulic stimulation is the process of injecting ﬂuid into a rock mass at or below the fracture opening pressure, and seeks to induce shear deformation on favorably oriented natural fractures [3]. In thermal stimulation, injectivity increases with injection time and with temperature contrast between the reservoir and cold injection temperature [6]. Chemical stimulation involves a mixture of acids that are injected into a well in order to dissolve material clogging the 6 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use fracture system of the reservoir, and aims to improve near- well permeability similar to that of hydraulic stimulation. In addition to the three methods described, other methods such as explosives are used in the oil and gas industry and are further explained in the oil and gas applications section 3.1.1.

2.1.1 Hydraulic stimulation

Hydraulic stimulation involves several mechanisms that are contributing factors to shear deformation due to the injection of fluid into a rock mass. The types of mechanisms described in this thesis are fluid mechanics, solid mechanics, fracture mechanics, and thermal mechanics. The composition of frac fluid also plays an important role in the type of fracture created, but each option possesses several advantages and disadvantages. To measure the effects of hydraulic stimulation, well tests are performed before, during, and/or after stimulation operations. Well tests include measurements of injectivity and productivity improvement ratios within a particular well.

2.1.1.1 Mechanics

The mechanics of hydraulic stimulation is divided into four categories. In stimulation (or fracturing), fluid mechanics describes the flow of one, two, or three phases within the fracture. Solid mechanics describes the deformation or opening of the rock because of the fluid pressure. Fracture mechanics describes all aspects of the failure and parting that occur near the tip of the hydraulic fracture. Thermal mechanics describes the exchange of heat between the fracturing fluid and the formation. Each case is governed by different factors that all play a role in hydraulic stimulation.

2.1.1.1.1 Fluid mechanics Fluid mechanics of a geothermal reservoir play an important role in well stimulation. Factors that are associated with ﬂuid mechanics in a geothermal reservoir include porosity, permeability, reservoir pressure, and skin effect.

Porosity (φ) is the fraction of total formation volume that is not occupied by solid rock (i.e., filled with formation fluid), measured as volume fraction (range from 0 to 1) or per- cent (from 0 to 100). Porosity is used in many correlations to develop a first estimate of other properties, such as rock strength or permeability, and relies mostly on the space present within particles at the time of deposition [7]. A distinction between total porosity

(φtotal) and effective porosity (φeff ) is noted as each governs assumptions for different types of flow. The total porosity is the volume not occupied by solid rock, but part of Cari Covell 7 the volume is occupied by bounded fluid that cannot move. The effective porosity is the volume occupied by moveable fluids, and it is the porosity of interest for most oilfield applications [7]. A notable exception is the use of φtotal for all reservoir calculations involving transient flow, or flow affected by changes in velocity and pressure [7]. In addition, no open or cased hole log measures porosity directly, but rather a property related to porosity is measured, such as density or resistivity. This is why a combination of porosity measurements is preferred for estimating φeff [7]. The most exact porosity measurements are made on cores, should they be available. Density tools are used to measure the elec- tron density of a formation, which is extremely close to its bulk density ρb [8]. If the density of the matrix components ρma and that of the pore fluid ρf are known, the total porosity from density can be found by volume balance:

ρma − ρb φD = (1) ρma − ρf where ρma is determined from the lithology and ρf is taken as that of the mud ﬁltrate, which is obtained from charts as a function of temperature, pressure and salinity [7].

Permeability is a measure of the ease with which fluids can flow through a formation, and the value of permeability depends on the orientation of flow. In the absence of natural fractures, the permeability kh parallel to the bedding of the formation is considered isotropic, or identical in all directions. However when natural fractures are present, kh varies and have a preferred direction. The permeability perpendicular to the bedding kv relates to the dominant geological nature of the reservoir and is usually only accounted for in horizontal wells [7]. Relative permeability accounts for more than one fluid type, and is sparsely considered in low temperature geothermal well stimulation [7]. Direct measurement of permeability can be obtained from well tests or sampling cores.

Reservoir pressure, or pore pressure, is the pressure of the fluid in a geological formation. Pore pressure is an input for designing stimulation treatments of multiple layers in the reservoir, in order to account for crossflow between zones. After production, its value can differ significantly from one layer to the next within a formation [7]. Reservoir pressure is obtained by a point measurement from well tests and the gaps between the measurements are filled by building pressure profiles.

Skin effect is a measure of the damage inflicted to the formation permeability in the vicinity of the wellbore. Damage may result from drill cuttings and mud cake from completion processes or from the production of formation fluids, and skin effect can therefore vary during the lifetime of a well [7]. For the design and execution of a stimulation treatment, the skin effect of interest is that related to the injection of treatment fluids into the for- 8 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use mation. The injection skin effect is obtained after conducting an injection test, explained further in section 2.1.1.3.

2.1.1.1.2 Solid mechanics Solid mechanics theory is associated with rock stress properties in an active geothermal area. Elastic and failure parameters are used in stress models to obtain a stress profile as a function of depth and rock properties. These profiles are important for estimating the stress variation between layers, and consequently the geometry of hydraulically induced fractures [7]. The parameters involved are Young’s modulus, Poisson’s ratio, and the poroelastic coefficient. Young’s modulus (E) defines the elastic relationship between stress and strain in a material. Poisson’s ratio (ν) is the fraction of rock expansion in the transverse direction divided by the fraction of rock compression in the axial direction. The poroelastic stress coefficient (η) controls the value of stress changes induced by pore pressure changes that result from depletion, injection or fracture fluid loss. All of these values are accounted for when designing well stimulation treatments, as they are associated with the rock type constructed in the applicable stimulation modeling software. Note that in stimulation (or fracturing) practices, seismicity is an effect of rock stress changes, which is discussed further in the Environmental impacts and seismicity section 2.3.

2.1.1.1.3 Fracture mechanics Fracture mechanics are important to analyze when modeling a stimulation treatment. Factors that are associated with fracture mechanics include tip effects and ﬂuid losses.

Fracture tip effects are governed by some net pressure in the fracture that controls fracture width. Net pressure and fracture height are directly related, and net pressure is dependent on fluid viscosity and pump rate. However in some cases, field observations have shown net pressure, and presumably fracture width, to be greater than predicted [9]. In such cases the fluid viscosity has a small effect on fracture width [9]. At a constant pump rate, it can be assumed that there is no net pressure at the fracture tip; i.e. fracture tip effects are ignored and is therefore a valid assumption to make when modeling a pre-stimulation treatment [7].

Fluid loss is a major fracture design variable characterized by a fluid loss coefficient C and a spurt loss coefficient Sp. The total fluid loss from the fracture (volume loss) is controlled by the total fluid loss coefficient C, where the volume lost while a hydraulic fracture treatment is being pumped can be approximated by [10][11]:

∼ p VLp = 6ChLL tp + 4LhLSp (2) Cari Covell 9 where

hL = permeable or ﬂuid loss height, L = fracture penetration length, and

tp = pumping time for a treatment.

The magnitude of C is typically from 0.0005 to 0.05 ft/min1/2 [7]. Spurt loss is the instantaneous volume loss of fluid per unit area that occurs prior to the development of a filter cake, and occurs only for wall-building fluids [7]. After stimulation, the fluid losses can be measured through field tests.

2.1.1.1.4 Thermal mechanics The properties of fracturing fluids show some depen- dence on temperature. In a typical fracturing treatment, the fluid is pumped at a temperature significantly below the reservoir temperature. As the fluid penetrates farther into the fracture, heat transfer occurs between the fracturing fluid and the rock, resulting in an increase in fluid temperature. The temperature gradient in the direction perpendicular to the fracture wall is significantly larger than those in other directions, so the temperature gradients in the other directions can be neglected [7]. In addition, heat conduction in the fluid can be ignored because it is small relative to both conduction in the rock and transport of heat with the moving fluid [7]. These assumptions reduce the heat transfer problem to a 1D problem perpendicular to the fracture wall, with conduction through the rock to the fracture face and convection from the rock face into the fluid.

2.1.1.2 Frac ﬂuid treatment

Hydraulic stimulation is usually conducted in two stages of fluid injection [12]. First, the pad stage is where only the hydraulic fracturing fluid, mainly water, is injected into the well to either clean out drill cuttings or breakdown the geological formation. Second, the slurry stage is where a mixture of fracturing fluid and propping solid material called a "proppant" is injected into the well and into the fractures. There are mainly three types of proppant injection fluid methods used during hydraulic stimulation within a pay zone. A pay zone is the portion of rock in a reservoir that contains economically producible hydrocarbons, or hot water in geothermal applications [13].

Hydraulic Proppant Fracturing (HPF) is the most conventional method in use for hydraulic stimulation [12]. HPF uses highly viscous gel as fracturing ﬂuid, usually in the form of a polymer. A high proppant concentration creates conductive yet relatively short 10 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use fractures in porous media suitable in reducing permeability impairments (i.e. "skin") in the wellbore, as illustrated in Figure1. The well is shut after the fracturing process to allow proppant transport through the fractures. However, HPF is prone to leave gel residues and may result in the precipitation of minerals, which affects well performance [12][14].

Water Fracturing (WF), or "Water Fracs", is essentially water containing friction-reducing agents added with a low proppant concentration. The WF method creates long and narrow fractures from the wellbore to the natural fracture network, which is at some distance from the main reservoir, as illustrated in Figure2. The fracture conductivity induced by WF is maintained by the self-propping ability of the reservoir rock. Since WF is dependent on the self-propping ability of the reservoir formation, fracture closure is likely to occur rapidly as a result of pressure solution processes in regions of high stress [12][14]. In addition, the low viscosity of water makes it difﬁcult to effectively transport proppants into the newly created hydraulic fractures [12].

Hybrid Fracturing (HF), or "Hybrid Fracs", is a combination of different gels used in the HPF method and slick water ﬂuids used in the WF method, or otherwise known as a cross-linked gel proppant. The concept is to utilize the advantages of the HPF and WF methods in creating the fracture geometry as well as effectively placing the proppant into the induced fracture. In the HF method, the fractures are considerably longer compared to HPF and the effective propped fracture length is higher compared to WF [14]. The HF method usually inherits the same problems as the parent frac methods [12]. Cari Covell 11

Figure 1: Fracture propagation as a result of Hydraulic Proppant Fracturing [14]

Figure 2: Fracture propagation as a result of Water Fracturing [14] 12 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

2.1.1.3 Hydraulic well testing

A series of hydraulic tests are performed before, during, and/or after a stimulation treatment. These include short or long term injection tests to measure reservoir characteristics, and production step tests for measuring well output. Well testing is performed to compare well behavior before and after a stimulation, and is typically used as a point of reference during stimulation between cycles.

The injection well test is a field test method where fresh water is injected into the well to raise the water level until a steady height is attained, and the pressure or water level change in the well is recorded. When a well is subjected to injection in order to monitor the pressure response in a reservoir (i.e. pressure transient), it is used to evaluate the properties that govern flow characteristics in the well. These properties include permeability, wellbore skin, storativity, transmissivity, initial pressure, and reservoir boundaries. Per- meability and wellbore skin have been previously described in the Fluid mechanics section 2.1.1.1.1. The storativity describes pressure movement within the reservoir, and is defined as:

S = cth (3) where

S = storativity [m3/Pa·m2], −1 ct = compressibility of the ﬂuid [Pa ], and h = effective reservoir thickness [m].

The transmissivity describes the ability of the reservoir to transmit ﬂuid, which mainly effects the pressure gradient between the well and the reservoir, and is deﬁned as:

kh T = (4) µ where

T = transmissivity [m3/Pa·s], k = permeability of the rock [m2], h = effective reservoir thickness [m], and µ = dynamic viscosity of the ﬂuid [Pa·s]. Cari Covell 13

During an injection test, the injectivity index is often used as an estimate of the connectivity of the well to the surrounding reservoir and is deﬁned as:

∆Q II = (5) ∆P where

II = injectivity index [(L/s)/bar], ∆Q = change in ﬂow rate [L/s], and ∆P = change in pressure [bar].

In the case of low-temperature wells tested through production step testing a comparable index is defined, termed productivity index (PI). The productivity index is a measure of well potential, or ability to produce, and is defined as the total mass flow rate per unit pressure drawdown, as explained further in the MProd governing equations section of Chapter 4: Methods.

The injectivity index (as well as the productivity index) is a simple relationship, approximately reflecting the capacity of a well, which is useful for determining whether a well is sufficiently open to be a successful producer and for comparison with other wells [15]. This neglects, however, transient changes and turbulence pressure drop at high flow-rates.

2.1.2 Thermal stimulation

Thermal stimulation relies on the thermal contraction induced by a significant temperature difference between the cold injection fluid against the hot rock formation (to create new fractures) and the enhancement of near wellbore permeability. Thermal cracking is attained by alternately injecting cold fluid and allowing the well to heat up the formation as thermal recovery ensues. Cold fluids may include cooling tower condensate, fresh water, seawater, or cold waste brine [12][16]. Several mechanisms that may enhance reservoir permeability include: 1) the reopening of pre-existing fractures due to thermally induced rock contraction, 2) the shearing of pre-existing fractures, 3) the creation of new fractures due to thermally induced stress changes, or 4) the development of secondary fractures due to the contrast in the thermo-elastic properties of rocks’ mineral components. Cleaning out of drilling cuttings that clog feed zones can also contribute to the effective permeability of the well, particularly in the initial cycles. 14 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

To date, a standard procedure for conducting thermal stimulation operations has not been established as the mechanism of cold water stimulation is still poorly understood [16][17] [18]. Such practices include cold ﬂuid injection through the drill string when the drilling rig is still on site, through an open ended drill pipe left in the well, or at a closed wellhead after the drill rig is removed [16][19]. In addition, thermal stimulation pumping pressures are usually kept relatively low so as to not cause hydraulic stimulation, yet productivity improvements have been achieved even if the warming stage has been excluded [12]. For the purposes of this thesis, thermal stimulation will not be discussed in great detail.

2.1.3 Chemical stimulation

Chemical stimulation, or more commonly known as acid stimulation, began as far back as 1895 when hydrochloric acid (HCl) was used to treat oil wells [20]. Despite success with HCl, it did not gain popularity due to corrosion problems affecting wellbore casings [12]. The use of acid was again attempted in 1932 when Grebe and Stoesser of Dow Chemical Co. discovered arsenic as a corrosion inhibitor. Acidizing technology has since advanced with the development of additives, methods, and systems to improve zone coverage during the acidizing process [7].

Chemical stimulation in the geothermal industry began in 1977 with the application of sodium carbonate (Na2CO3) in the Fenton Hill Hot Dry Rock (HDR) project in New Mexico, USA in attempts to reduce flow impedance [21]. Further programs throughout the 1980s in the Geysers geothermal field of California were designed to create new conductive flow paths to the main reservoir. Shortly afterwards, acid stimulation gained popularity in places such as Central America, Philippines, and Italy with improvements in injectivity ranging from 40% to over 300% increase from original values [12].

Throughout the geothermal industry, there are two main methods of acid stimulation that are in practice today. Matrix acidizing involves acid treatment injected at pressures below the formation fracturing pressure and is designed to remove skin or other formations of damage that may occur during well operation [20]. Acid fracturing or "fracture acidizing" is designed to stimulate an undamaged formation and is conducted above the formation fracturing pressure [12]. While chemical stimulation treatments are of great value to the geothermal industry, they are not of direct concern for the purposes of this thesis and therefore will not be discussed in more detail. Cari Covell 15

2.2 Types of hydraulic stimulation

The methods of hydraulic stimulation are described in three categories, as applied on geothermal wells. Air-lift aided drilling (also referred to as pressure balanced drilling, under-balanced drilling (UBD), or aerated drilling) has proven to be successful in pre- venting the clogging up of feed zones during drilling, but is not necessarily considered a stimulation operation per se [19]. However air-lift pumping that is followed by water circulation helps to restore feed zone permeability that was possibly reduced during drilling [19]. The water circulation phase could be classiﬁed as an open-hole stimulation via injection at the wellhead, and is performed in a similar matter. Another method of stimulation is the isolation of intervals in the borehole through the use of a packer in particular to an open hole section. After the packer is set, water may be injected either below the packer, through the drill pipe, or into the annulus above the packer [22]. By using a packer for zonal isolation, a larger effective fracture area can be obtained rather than one massive stimulation over a long open hole section. The packer method is also favorable to reduce the risk of creating larger seismic events [23]. Double packers are also considered to be a method of hydraulic stimulation, but the method is not discussed in this thesis. This is because double packers have hardly been used in geothermal stimulation operations; even though they are potentially more powerful than a single packer due to injected water being focused within a shorter interval [19]. Lastly, zonal isolation is a method used to target one fracture at a time within the wellbore. Several other options are available to achieve zonal isolation, where each technique must be integrated into drilling and well construction [24].

2.2.1 Air-lift pumping

Air-lift testing aids in maximizing well output and is therefore deﬁned as a valid stimulation technique for the basis of this thesis. Typically air-lift testing is done with compressed air or nitrogen gas, although other gases are known to have been used in geothermal applications that work better with natural gas production of the reservoir [25]. During injection, lift gas is compressed to a pressure equal to or greater than reservoir pressure at the depth of the lower end of the air line (drill pipe). A schematic diagram for air-lift drilling (pressure balance) stimulation is shown in Figure3.

The design for air-lift pumping is based on rules of thumb and graphs that have been developed for air lifting freshwater wells. The main parameters are the submergence of the airline (the coiled tubing) and the air volume and pressure required. The design basis 16 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use can be found in the book “Groundwater and Wells” by Driscol [26] and the Australian training manual "Drilling: The Manual of Methods, Applications and Management" [27]. Figure4 shows the basis for designing the air lifting program.

Figure 3: Schematic illustration of the setup for air-lift pumping [19] Cari Covell 17

Figure 4: Diagram for design of air-lift pumping, based on water well experience [26]

2.2.2 Open-hole with a packer

The procedure to run an open-hole packer stimulation is based on the full Icelandic paper from Axelsson et al. [19]. The paper explains the open-hole packer procedure using a Baker Hughes - Baker Oil Tools packer of 10.375" OD SS OH ISP Bridge Plug for a well of approximately 2,000 m depth. A more detailed cross section of the packer can be seen in AppendixA. Figure5 shows the general schematic of the open-hole packer procedure. Note that injection is done either below the packer or above the packer, typically one after the other, for any given stimulation. The step-by-step procedure for implementation, stimulation, and removal of the packer is described in the following sections. 18 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Figure 5: Schematic picture of injection via a packer [22]

2.2.2.1 Before deploying the packer

The well is first filled to the brim with water in order to measure fluid loss by monitoring the pump rate. Once a steady pump rate has been determined, hold constant for about 1.5 hours and monitor the pressure on the wellhead ("kill line"). A logger is then attached to the mandrel below the packer, where temperature and pressure data will be measured in 3 second intervals. For the logger, it is important to keep the well cooled by constant pumping of cold water. The Icelandic report indicates that pumping should not be stopped for more than 15 minutes due to temperature rise in the well while there is no circulation. The packer is then lifted upright and oriented so that the bleeding cap is facing upwards. The bleeding cap is opened to release any air trapped or excess water. Should water not come out of the bleed valve, the packer will need to be completely filled with water. Air should not enter the packer after this procedure. Pressure tests are then run on all surface Cari Covell 19 piping from the pumps to the rig floor. Lastly, the packer is connected to the drill pipe and prepared to run into the hole.

2.2.2.2 Installing the packer

As the packer is installed, water injection via the kill line is maintained until the packer reaches approximately 690 m depth inside the production casing. There is no need to pump through the drill string until 1000 m depth is reached because formation of the geothermal ﬁeld is typically not very hot above this depth. The logger will show depth in meters with respect to the top of the packer, therefore it is necessary to account for the depth of each joint on the drill pipe in order to measure the true depth. The running speed of the packer is about 30 seconds for every 12.5 m; or the length between each drill pipe joint. Assuming a water level of 150 m depth, the pressure on the drill string should not exceed 15 bar in order to avoid premature inﬂation of the packer. The pressure on the drill string should not exceed 40 bar if the hole is assumed to be full of water.

2.2.2.3 Setting the packer

A single-set packer can only be set one time per stimulation treatment interval and therefore must be pulled out after each test. Multi-set packers with different setting valves are also available. Typically, the packer is set at the deepest interval first and subsequently moved up to shallower intervals. For setting in directional wells, the packer is normally placed within a 10 m range of the desired interval for stimulation, where there is slight tension once the setting depth is reached. A chalk mark on the drill pipe is made above the rig floor to mark the exact depth setting position and the hook load weight is subsequently determined. The drill pipe connection is then opened to drop the 1.5" ball, where the ball will reach the shear plug at the falling speed of about 300 meters in 5 minutes. To check for any leaks on the surface, the drill string is filled by injecting water slowly until a 15-20 bar pressure is reached for a water level of 150 m; or about 34 bar if the hole is filled. The pump pressure is then increased in increments of 20 bar until a pressure of 80 bar is reached for the 150 m water level scenario; or 95 bar if filled with water. The pressure is maintained for approximately 15 minutes to ensure that the packer is fully inflated, then released rapidly in order to get the swift closure of the check valve that maintains the inflation of the packer. Assuming a 12 1/4" diameter hole, a weight of 4.5 tons is applied to the drill string to check that the packer is firmly anchored in the well. 20 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

2.2.2.4 Opening the bottom plug

The opening of the bottom plug allows water to flow through the packer. Pressure is then increased to about 107 bar for water depth of 150 m, or 122 bar for the filled hole, in order to break the shear pin and release the bottom plug and ball. Subsequently, the plug falls to the bottom of the well. The well is then filled via the kill line at a rate of about 15 L/s to measure the natural fluid loss in the open hole section above the packer.

2.2.2.5 Stimulation

Stimulation begins by slowly increasing the pump rate to reach the capacity of three rig pumps, and is maintained for about 30 minutes. The pressure should remain steady during this period. Pumping should then be stopped for 15 minutes after 1.5 hours of continuous injection to observe the pressure falloff. The intermittent no-pumping time is considered beneficial to the stimulation process [23]. In addition, the water supply system and rig mud tank capacity may not be great enough to keep up with the large amount of water being pumped into the well and injection will need to stop to refill the tanks. Information about pressure and flow should be hand recorded as backup to the down-hole and on surface digital data records. The method of increasing pressure and stopping the pumping process in the specified intervals is usually maintained for a period of 12-24 hours.

2.2.2.6 Releasing the packer

The packer is first deflated by a slight clockwise torque applied to the drill string. A 2.25 ton weight is placed over the string for about 10 minutes to allow for the rubber to deflate completely. The packer should then be free to pull out of the hole. Final steps include circulating cold water for logger retrieval.

2.2.3 Zonal isolation

The zonal isolation discussion in this thesis will be guided under four categories. A simple approach is to use a liner made out of cement or sand that is perforated to distribute fluid across the reservoir. The addition of an inflatable or swellable packer creates a seal to isolate the desired stimulation interval and is an option available for use in the field. The plug and go method involves drilling until a fracture is reached, stimulating the fracture, and then isolating the interval to be repeated over the entire depth of the wellbore. Finally, the use of multilateral wellbores and sidetracks involves first drilling a pilot hole, then Cari Covell 21 drilling either multiple holes to intersect individual fractures or drilling a deviation hole to more accurately target the upper fractures of the reservoir.

2.2.3.1 Use of a liner

In order to provide zonal isolation between fracture zones, one option is to run a liner that is placed with cement. A cement liner is typically used as this option does not disrupt drilling operations during well construction and does not compromise the wellbore shape and geometry [24]. The liner is initially set above the bottom fracture, perforated to provide hydraulic access to the reservoir, and then stimulated in open hole [24]. A plug or isolation packer is then set in the liner and another stimulation is performed to target the fracture. To access the next fracture, the plug or isolation packer is pulled and reset above the stimulated fracture. The liner is perforated and the fracture is stimulated at the desired interval, and the process is repeated until the final interval is reached. Note that experience proves the use of a cement liner to significantly impair the permeability of the fractures, as it may be difficult to achieve sufficient injectivity from the well [24]. However, viable solutions for extreme high temperatures, lost circulation, CO2 attack, and cement/casing integrity have been tested for safe geothermal operation [28].

As an addition to the cemented liner, stage cementing collars are optional materials to implement in stimulation practices. The collars are staged based on previously logged spacing between the target fracture zones. A wiper plug is pumped to land at the shoe of the liner, followed by a dart that lands in the lowermost stage cement collar located above the second fracture zone, and an open sleeve is observed after application of pressure from surface shifts [24]. The collar displaces the next volume of cement to form an annular seal above the second fracture zone. The process is then repeated for the next stage collar. Careful observation of the pumping pressures and displacement volumes is necessary as the parameters are sensitive relative to the cement used in particular to high temperature geothermal systems [24].

A sanded liner is also an option for zonal isolation, where sand is pumped into the annular space so that the liner is supported [24]. Stimulation is performed in the same manner as a cement liner. However, the sand liner has a number of disadvantages. Given the anticipated nature and scale of the fracture openings, the size of sand particles will not adequately bridge off across the fracture. Effects include some sand being lost in the fractures, losses in permeability and injectivity, and an increased risk of stimulating pressures to areas not in the target zone [24]. 22 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

2.2.3.2 Use of a liner with inﬂatable or swellable packers

External Casing Packers (ECPs) are located for zonal isolation between multiple fracture zones. ECPs are a type of packer on the outside of the casing or liner that is inﬂated with drilling mud or cement. The packer inﬂates to conform to the wellbore by forming a seal to the liner and subsequently isolates the zone. The stimulation may then be completed using several options including: sliding sleeves, plugs and perforating tools, stage collars above the ECP, or straddle packer assemblies across the required intervals [24].

Swellable packers are comprised of an elastomer element that swells naturally when ex- posed to the appropriate swelling agent that is either water or oil based ﬂuid [24]. The packer is bonded to the outside of the base pipe and is assembled for inclusion in the liner. Once the desired depth is located, the well is displaced to a water based ﬂuid, and then time is required to allow the packers to swell to ensure a seal against the wellbore wall [24]. As a note, the packer is recommended for use when there is good control of wellbore geometry as swellable elements are limited in the extent of expansion, and as a result the differential pressure that can be applied is limited [24].

2.2.3.3 Plug and go

In the plug and go method, drilling is ﬁrst done to the uppermost fracture and stimulation is performed by setting a packer in the production casing. For the isolation phase, two methods are known to be effective. One option is to install an expandable liner across the fracture zone and tie back to the production casing. A seal is made by either using cement or swellable packers on the outside of the expandable liner [24]. Drilling then continues to the next fracture zone, where stimulation and placement of a liner is repeated. In this case, the liner can be tied back to the pre-existing liner or be simply set across the target fracture zone [24]. The process is repeated until the entire wellbore section is drilled and stimulated. While this method has been proven to be effective, it may be desirable to limit the number of well re-entries for operational purposes [24]. Another option for isolation is to install a conventional liner with swellable packers. The elastomer in the packer absorbs the formation water over a period of time and swells to form a seal between the liner and the wellbore [24]. Currently the technology is being developed to handle high temperatures of around 300°C [24]. Cari Covell 23

2.2.3.4 Multilateral wellbores and sidetracks

The multilateral wellbore method comes about after the bottom fracture zone has been stimulated, isolated with a packer, and cased off above that fracture. The depths of the upper fractures measured during the drilling of the pilot hole are used as target locations for drilling additional wells laterally. All lateral wells are plugged prior to drilling the next to ensure isolation for drilling and stimulation purposes [24]. Multilateral wellbores are efﬁcient in accessing each leg separately, but costs add up for each wellbore and the multilateral technology relies heavily on the ability of elastomer seals to handle high temperatures [24]. The use of sidetracks is considered another option for better targeting the upper fracture zones. A down-hole tool with an inclined plane called a whipstock is placed in the wellbore to exit the original hole and to drill a new well to the next intersected fracture. The fracture is stimulated and isolated, and the process continues until all desired fracture zones have been targeted. The use of sidetracks as an option for zonal isolation is also effective, but it relies almost exclusively on the ability to drill successfully to each interval [24].

2.3 Environmental impacts and seismicity

Besides the technical challenges to stimulate fractures along a fault system, there are other environmental challenges to consider. These include the use of chemicals in the fracking ﬂuid, drilling noise, damage to ﬂora, and changes in thermal manifestations. Induced seismicity has also received much attention after seismic events following hydraulic stimulation in Australia, France, and Switzerland have been the topic of many studies.

In some low enthalpy geothermal fields, impact from long-term utilization of water may include the dropping water level of near surface aquifers. In addition, flow reduction or dry-up of nearby springs and shallow water wells may occur. All these problems can be avoided by reinjecting the cooled liquid via stimulation practices [23]. The chemistry of fracking fluid is also of environmental concern, as acid behavior in a reservoir during chemical stimulation can cause scaling due to an improper mixture of tracers. Some proppants within the injection fluid consist of ceramic or glass beads and could be coated or constructed using reactive metals, although this is rare in geothermal applications [29]. To address these environmental concerns, there are groundwater treatment solutions that can be implemented after fracking fluids have been injected; such as reactant sand-fracking (RSF) to decrease metal contamination in aquifers and permeable reactive barriers (PRB) to decrease the amount of volatile organic compounds [29]. 24 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Noise can be severe and a nuisance to local residents who live close to the geothermal plant, but is countered by proper plant engineering (e.g. avoiding noisy equipment) and by placing noise barriers if necessary [23]. Damage to local flora occurs every time steam or hot geothermal fluids are released on local plants, which happens due to the high temperature and salinity of the geothermal fluids, but is usually not a problem when fluids are reinjected [23]. Some changes in thermal manifestations are possible during exploitation, where hot spring and fumarole intensity depend on the pressure drawdown of the reservoir. Hydrothermal eruptions could also be correlated with exploitation, but only one such case in New Zealand has occurred [23].

The most problematic side-effect of enhancing geothermal reservoirs by hydraulic stimulation is the potential to generate earthquakes, which may become detrimental for further development of project [23]. During hydraulic stimulation, stress patterns in the rock change to potentially cause microseismic events, but most events are of very low magnitude and are short in duration. However some EGS sites have recorded significant seismic events arising from short-term stimulation operations. In Cooper Basin, Australia, more than 32,000 seismic events occurred in well Habanero-1 after stimulation via large volume injection of 20,000 m3 of water [30]. The largest seismic event recorded was of a 3.7 magnitude. Similarly, the Soultz-sous-Forêts project involved several stimulations that averaged 23,000 m3 of water injected and thousands of microseismic events were induced each time, where the largest event reached 2.9 magnitude [31]. The stimulation program had to be changed due to complaints from local residents [23][30]. In Basil, Switzerland, rupture processes for seismic events with a magnitude greater than 2.2 appear to have occurred in cascades either on a single continuous fracture or nearly synchronously on several closely adjacent structures [32]. Due to this, stimulation operations were put on preliminary halt [23]. The size and frequency of seismic events are therefore primarily controlled by the rate and amount of fluid injected, but are also controlled by the orientation of the stress field [30][33][34][35][36]. 25

Chapter 3

Literature review

Stimulation practices originated, and are still in use, in the oil and gas industry and have translated to geothermal applications beginning in the early 1970s. When comparing well stimulation applications between the oil and gas industry and the geothermal industry quantitatively, geothermal well stimulation only accounts for a small fraction of all cases. This is mainly due to the fact that geothermal exploration is a younger industry, and most stimulations are ones that include specialized experiments for research rather than for commercial use. The literature review conducted in this chapter will discuss several cases of hydraulic stimulation for each ﬁeld type in order to evaluate potential correlation.

3.1 Hydraulic stimulation applications

The earliest form of hydraulic stimulation in the oil and gas industry is approximately around the late 1800s-early 1900s [25]. This section will focus on hydraulic stimulation applications because more commercial work has been performed in the oil and gas industry that has translated well to the geothermal industry regarding experiences and lessons learned. Thermal and chemical stimulation are still relatively new in the geothermal industry and most projects are research oriented within high-temperature geothermal ﬁelds. EGS is important to discuss as they have been the center of study in hydraulic stimulation since the late 1970s. 26 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

3.1.1 Oil and gas industry

Stimulation techniques to improve well permeability have been in practice within the oil and gas industry since the late 1800s. Examples of stimulation techniques include casing perforation, explosive propellant solution, acoustic stimulation, and electric stimulation. Casing perforation is designed to access cased-off permeable horizons in production wells that still have high commercial temperatures and pressures, and is one of the most common methods used in the oil and gas industry [37]. Horizons are typically found at the shallow depths of the reservoir, where further evaluation is performed based on drilling circulation losses, geology, and petrology of the formation prior to the conduct of a perforation operation [12]. High energy gas fracturing (HEGF), or explosive stimulation, creates a breakdown of the formation and at the same time improves clean up of the perforations [38]. High gas waves generated from the vaporizing propellant (termed deflagration) crushes the formation damage to create small fractures near the perforation channel. When pressure disperses, the gas creates a flooding effect and carries back the fine particles from the formation. Acoustic stimulation uses a simple ultrasonic wave source between the acoustic field and the saturated porous rock. In geothermal applications, this interaction changes permeability or removes plugging materials in the formation [12]. Electric stimulation uses electric current either through electrothermal or electrodynamic type effects. Electrothermal effect is evident in the near wellbore zone during heating with infared or high frequency microwaves, while electrodynamic effect creates a cleaning of the bottom hole formation zone from clay particles restoring or improving the well permeability [39]. While some of these techniques have translated to the geothermal industry, most of these techniques are in the novel stages of development and require more research before being tested on a wider scale [12].

3.1.2 Low temperature geothermal areas

Hydraulic stimulation by use of open-hole packers and large flow of water has had success in low temperature geothermal areas, specifically in Iceland starting in the early 1970s. The Reykir hydrothermal system was an ambitious stimulation program, when each of the 39 wells drilled during redevelopment of the field were stimulated after drilling. Selt- jarnarnes well SN-12 was drilled in 1994 after five previously drilled wells were used for exploration [40]. The decision to stimulate the well resulted from a measured flow yielding almost no production after drilling [41]. Each field and their corresponding stimulation programs are discussed in more detail in the subsequent sections. Cari Covell 27

Other low-temperature geothermal fields that have experienced well stimulation in Ice- land include Hlídardalur and surrounding fields in SW-Iceland, Laugarland and Leirá in N-Iceland, and Urridavatn in E-Iceland. Information is limited on these subjects, but reported findings are discussed in section 3.1.2.3.

3.1.2.1 Reykir hydrothermal system

The Reykir hydrothermal system has been exploited since 1944 for space heating of Reyk- javik, Iceland. Prior to 1970, production amounted to 300 l/s at 86°C by free flow from 69 wells [42]. In the early 1970’s, the Reykir field was redeveloped with the addition of 39 wells to be hydraulically stimulated, by injection above and below inflatable packers, as part of a drill completion program.

In general, air-lift pumping was done to clean the hole of drill cuttings and lost circulation materials. A packer was then set a certain depth, between two or more producing horizons, where water was injected beneath or above the packer. Pumping rate varied from 15 to 100 l/s for each well due to the resistance of the producing horizons. Pressure increases at the feed zones ranged from a few bars up to as high as 150 bars at the lowest permeability feed zones treated [42][4]. Tomasson [22] and Tomasson and Thorsteinsson [43] describe wells MG-25, MG-27, MG-35, and MG-39 in more detail in order to better illustrate the hydraulic stimulation method, discussed in the subsequent sections.

The outcomes described in the following are an average of all wells stimulated. Com- paring the injectivity before the stimulation and after indicated as much as 30-40 fold increase; so in total more than 1500 l/s was produced. The drastic improvement is mostly attributed to the reopening of feed-zones clogged by drill cuttings during drilling operation [4]. However, when comparing production over the cumulative loss of circulation during drilling, the wells showed a three-fold increase in production. This is attributed to increased feed-zone permeability, most likely due to the removal of zeolite and calcite vein deposits and partly to increased permeability of near-well fractures in hydroclastic rocks [4][43]. Speciﬁc results per well are referred to by [42].

3.1.2.1.1 Well MG-25 The well was drilled to a depth of 2025 m with a ﬂow of about

3 l/s in 1974. Circulation losses after drilling amounted to Q1 = 2 l/s and total circulation losses amounted to Q2 = 13 l/s given the static water level at 20 m [22]. The ﬁrst packer setting was at 758 m depth in dolerite intrusions with the biggest aquifer above the packer and many small aquifers beneath the packer as seen from lithologic logs of the well [22]. Dolerite tends to be more permeable than basalt and hyaloclastic rocks, therefore it is an 28 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use ideal location for stimulation in typical Icelandic geological environments. One run was done below the packer from 758-2025 m depth, and two runs were done above the packer from 204-758 m depth. The ﬁrst run had pressure increasing step-wise, while the two runs above the packer at different injection rates had constant pressure of 20 kg/cm2 and no pressure fall off [22].

After a short break, the ﬁrst interval of 758-2025 m was run three more times at constant injection rates of 37-38 l/s. A new packer setting at 552 m depth was then implemented with injection below the packer to the bottom of the drillhole in two runs at constant injection of 45-46 l/s. Both intervals experienced a drop in pressure, and gradually the pressure built up again meaning an opening of an aquifer most likely occurred [22].

The step draw-down test conducted after stimulation indicated a flow of 40 l/s with a 40 m drawdown [22]. The results show a 14 fold increase compared to fluid losses after drilling, and a 2.2 fold increase including fluid losses during during. This shows a great improvement in production and proves that drilling deeper drillholes can be beneficial for good producing horizons within the geothermal system [22].

3.1.2.1.2 Well MG-27 In September 1974, well MG-27 was air-lift pumped with compressed air for 12 hours with an estimated yield of 15 l/s. An injection test was run thereafter where total circulation losses were computed. These values were then used to calculate improvement ratios, of which the square roots indicate increase in productivity [43].

The injection packer was sequentially set at 1217 m, 951 m, and 835 m. Water was injected above and below the packer and multiple injection tests were made after each setting. The coefficient of turbulent loss C was calculated for each setting, which is a factor associated with natural fluid loss in the well due to turbulence [43]. Subsequently the improvement ratios were tabulated using values of C1 and C2; coefficients of turbulent losses during and after drilling. The tabulations are presented in Table1, along with the volumes withdrawn by the compressed air pumping and injections above and below the packer. Cari Covell 29

Table 1: Improvement ratios in MG-27 [43]

3 1/2 1/2 Depth [m] Volume injected [m ]CI1 I2 1/2 beneath above total [m/(l/s) ] [C1/C] [C2/C] Compr. air 650 0.250 2.16 0.35 1217 1135 980 2115 0.037 5.62 0.93 951 2350 0 2350 0.062 4.34 0.72 835 620 2600 3220 0.025 6.84 1.13

The high value of C = 0.062 computed after the 951 m setting is probably due to a tight spot and a partial cave-in at a depth of 270-320 m during the setting [43]. To avoid problems for subsequent stimulation at 835 m depth, the section was widened and the well was cleaned to the bottom with a drillbit. The most successful stimulation, in terms of increase in productivity, is at the 835 m depth with a 2.6 fold increase and a 1.06 fold increase taking into account circulation losses at the end of drilling and during drilling, respectively.

3.1.2.1.3 Well MG-35 In August 1976, well MG-35 was air-lift pumped with compressed air at a rate of 50-70 l/s with the static water level remaining at a depth of 80 m. The increase in temperature was measured with time to see how cooling points coincide with increases in turbidity; meaning that new and deeper zones of lost circulation are being cleaned out.

The packer was then set at a depth of 560 m and water was injected below the packer. The coefﬁcient of turbulent well losses after injection was computed as C = 0.022 m/(l/s)1/2 from the initial build-up pressure beneath the packer by the air pumping; because only one small loss of circulation occurred above the setting depth [43]. At the end of the stimulation, C had reduced to 0.019 m/(l/s)1/2 (see Table2).

Two additional packer settings were made at 1153 m and 1359 m where water was injected above and beneath the packer. The total losses of circulation at the end of drilling and during drilling were Q1 = 85 l/s and Q2 = 7 l/s, respectfully. The turbulent well loss 1/2 1/2 coefﬁcients are subsequently C1 = 1.63 m/(l/s) and C2 = 0.011 m/(l/s) . By using the coefﬁcient of turbulent well loss values, the improvement ratios were calculated and the end of drilling and during drilling and are presented in Table2. 30 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Table 2: Improvement ratios in MG-35 (based on [43])

1/2 1/2 Injection rate [l/s] C I1 I2 Depth [m] 1/2 above below [m/(l/s) ] [C1/C] [C2/C] Compr. air 0.022 74.09 0.5 50-70 560 0.019 85.79 0.58 1153 60 65 0.0015 1896 7.3 1359 40

At the end of stimulation, the productivity of the well increased by 32.9 fold after drilling, but only by 2.7 fold compared to total productivity from circulation losses during drilling

(numbers are the square root of I1 and I2). Well MG-35 is considered to be one of the best producing wells after stimulation, however such a large margin of productivity improvements may be due to other perpetuating factors of the geothermal ﬁeld that have an effect on circulation losses.

3.1.2.1.4 Well MG-39 In June 1977, well MG-39 was stimulated by injection above packers for two different depths at 1138 m and 1001 m. Injection rates were between 83-91 l/s with a known static water level of 93 m, but continuous injection was limited due to insufﬁcient water supply. Therefore, injection periods were limited to 15-35 min [43].

At the ﬁrst setting of 1138 m, pressure fell gradually from 10 kg/cm2 to 0 in two days, but at the 1001 m setting the pressure drop was from 7.5 to 3.0 kg/cm2 in 30 hours [43]. The total loss of circulation was 287 l/s, of which 49% occurred above the ﬁrst packer setting at 1138 m and 35% occurred above the second packer setting at 1001 m [43].

The total increase in productivity for well MG-39 was estimated to be 3.9 fold, not including circulation losses during drilling. After computing all turbulent losses, the increase in productivity was estimated to be 1.5 fold for the 1138 m packer depth and 1.2 fold for the 1001 m packer depth [43]. Hence, there was a 0.3 fold increase from stimulating 137 m deeper, so there may be a correlation of productivity with stimulation depth.

3.1.2.2 Seltjarnarnes well SN-12

The Seltjarnarnes geothermal ﬁeld has been exploited for hot water since 1970 to serve the town of Seltjarnarnes in SW-Iceland. Seltjarnarnes is a typical low temperature geothermal ﬁeld with reservoir temperatures ranging from 80 to over 140°C at 2700 m depth [44]. Cari Covell 31

The system consists of 3-4 different aquifers, with different temperatures and salinity. Supersaturation of calcium carbonate, due to mixing of water from different feed-zones within a well, has increased with time but no scaling has yet to occur. Although, calcium carbonate is now at a level where scaling is known to have occurred in other geothermal areas of Iceland [41].

Well SN-12 was drilled to a depth of 2714 m in the fall of 1994 and appeared to be almost non-productive after drilling; yielding ﬂow less than 1.5 l/s with a 150 m draw down after a one hour air-lift pumping test. Therefore, it was decided to stimulate the well with a packer in two phases over a ten day period [41]. Prior to stimulation, the average yearly production of the Seltjarnarnes geothermal ﬁeld was around 30 l/s since 1991 [41].

3.1.2.2.1 High-pressure wellhead injection The ﬁrst stimulation phase involved high- pressure wellhead injection of cold water. Water was injected at 60 l/s in one hour periods, because not enough water was available at the drill site and storage tanks were reﬁlled during 30 min breaks. At the end of the twelve hour pumping period, the wellhead pressure dropped suddenly from 76 bars to about 18 bars, indicating that the well had in fact been stimulated.

The wellhead injection continued for 12 hours and additional pressure drops were observed. At about 8 a.m. the pressure dropped suddenly by about 18 bars and at about 10 a.m. the wellhead pressure was down to 23 bars [41]. This indicated that the well had been stimulated even further, but the pressure started to increase towards the end of the injection phase and it became evident that the well had collapsed [41].

Based on water level monitoring of well SN-6, the transmissivity was estimated to equal T = 6.3×10−8 m3/Pa·s, and the storage coefﬁcient to be S = 3.5×10−9 m3/Pa , which correspond to a permeability of 15 mD [41]. The transmissivity may be compared to older estimates which are in the range of 3.2×10−8 m3/Pa·s to 40×10−8 m3/Pa·s [44]. The storage coefﬁcient is small, indicating that permeability is limited to a thin fracture- zone, perhaps on the order of 50 - 100 m [41].

3.1.2.2.2 High-pressure injection below packer The second stimulation phase involved high-pressure injection below a packer at 1412 m depth, in order to stimulate the lower part of the well further and to clean out feed-zones clogged by drill-cuttings [41]. The depth of the packer was chosen on the basis of temperature logs, caliper logs, and borehole lithology, which indicate the existence of feed-zones. 32 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Similar to the ﬁrst stimulation phase, 58-59 l/s were injected below the packer during four one hour intervals, each followed by a 20 min break. Accounting for pressure loss in the drill string, the pressure below the packer decreased gradually from 85 bars at the beginning down to 40 bars. Therefore, it is believed that the stimulation was due to the removal of drill cuttings clogging feed-zones [41].

3.1.2.2.3 Production testing Several production tests were performed throughout the stimulation program. A step-rate injection test was performed on the 13th of October 1994 after the ﬁrst phase of stimulation, and an air-lift test was performed after the second phase of stimulation from the 18th-19th of October 1994 (see Figure6). In addition, an air-lift test was performed at the end of drilling on the 20th of October 1994. The air-lift test after the second stimulation was done in four steps with ﬂow rate varying from 12 l/s to almost 30 l/s [41]. As seen in Figure6, the pressure drawdown varied from about 1.5 bars to almost 6 bars, making clear that the productivity of the well had improved.

Figure 6: Results of production testing of well SN-12, where symbols show observed data one hour into each step and lines show calculated output characteristics [41]

Comparing results from the air-lift tests indicate an increase in ﬂow to about 35 l/s with a draw down of roughly 60 m, and the stimulation had increased the yield of the well by a factor of 60. Thus well SN-12, which appeared to be almost non-productive at the completion of drilling, had turned into a good production well [4]. Cari Covell 33

3.1.2.3 Other low-temperature ﬁelds in Iceland

Several other low-temperature fields were stimulated in Iceland throughout the 1970s. The first packer experiment took place at Hlídardalur 50 km Southeast of Reykjavik, where production of a 1220 m deep narrow gage drillhole was increased from 1 l/s, 60°C, with a drawdown of 100 m, to 2-3 l/s, 100°C, by free flow [22]. While 2-3 l/s is not much flow for a typical low-temperature geothermal well, the increase in temperature shows that the well was enhanced and was sufficient for the boarding school, swimming pool and adjacent buildings.

Good results were achieved in three other drillholes ranging from 1097-1733 m depth, located in Selfoss, Midsandur, and Thoroddstadir, SW-Iceland [43]. However at Litla- land near Hlídardalur, a 2187 m deep well collapsed after a packer set at 576 m depth experienced pressures of 70-80 kg/cm2 below the packer [43].

In Laugarland, N-Iceland, there were no improvements detected in stimulation of wells LJ-6 and LJ-8 because pressures as high as 130-150 kg/cm2 were experienced due to multiple collapses of the drillhole at depths below 940 m [43][45]. Flows were less than 1 l/s and came mostly from shallow veins around 250 m depth, even after injection at rates between 30-35 l/s [45].

Some improvements were seen in Tertiary rock, such as in Leirá (SW-Iceland) in the uppermost 500 m section of a 2019 m well, and in Siglufjördur (N-Iceland) the productivity of a 1100 m deep well increased about 50% after several packer settings [43]. While improvements were shown in these areas, the margin of success seems small compared to the time spent stimulating the well at multiple packer depths, therefore drillhole geology and the effects of ﬂuid losses may be underlying mechanisms that directly inﬂuence production output.

At Urridavatn only small improvements were seen below 250 m while 2-3 fold improvements were seen between 172-250 m [43]. The results from Urridavatn show high sensitivity regarding the interval at which stimulation succeeds; which questions the speciﬁc geological environment required for stimulation location.

3.1.3 High temperature geothermal areas

The first hydraulic stimulation experiments in high temperature geothermal areas were performed from 1978-1983 at the Nigorikawa and Kakkonda geothermal fields in Japan, but detailed information on the stimulation programs is not available at this time [46]. The 34 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use earliest form of packer and proppant technology in high temperature geothermal areas was when the Department of Energy (DOE) implemented a stimulation program in the United States throughout the early 1980s. Around the same time, hydraulic stimulation through wellhead injection occurred in Italy as part of a different stimulation program set up by the National Entity for Electricity (ENEL). Due to the other thermal and chemical stimulation research occurring throughout these programs, as well as other EGS programs occurring simultaneously, one could say there was a gap in time of no hydraulic stimulation in high temperature geothermal areas, at least according to the definitions of high temperature and EGS fields previously discussed in the introduction of this thesis. Hydraulic stimulation via wellhead injection then resumed in the late 2000s with the Salak field of Indonesia and the Mt. Apo area of the Philippines in order to increase the permeability of these regions. From 2012-2013, the Salak field did a zonal isolation experiment in order to prove the viability of such a stimulation experiment [47].

However, hydraulic stimulation in high temperature areas can be difﬁcult due to several factors of the high temperature nature of geothermal areas effecting the potential for production improvements. Therefore, stimulation in high temperature geothermal wells is more common through thermal injection of cold water and allowing the well to heat up. Several examples of thermal stimulation in high temperature areas around Iceland are available for review. Each case of thermal stimulation in Iceland is discussed further in AppendixB in order to analyze effects based on the unique geology and how these effects inﬂuence production outcomes of particular wells.

3.1.3.1 Baca, New Mexico (USA)

Hydraulic stimulation in the Baca geothermal ﬁeld of New Mexico, USA was done in 1981 on wells Baca 23 and Baca 20 as part of the DOE-sponsored Geothermal Reservoir Well Stimulation Program (GRWSP) [48][49]. The liquid dominated reservoir is com- posed of volcanic tuffs with low permeability and temperatures as high as 500°F (260°C) [49].

3.1.3.1.1 Well Baca 23 The well was originally completed with a cemented liner at 3057 feet and open hole to 5700 feet with an interval from 3300 feet to 3500 feet selected for hydraulic stimulation. The section was isolated using an experimental high temperature Otis packer with ethylene propylene diene monomer (EDPM) elastomer elements and was treated using a combination of ’frac ﬂuid’ water with a ﬂuid loss additive (FLA) and polymer gel proppants [48]. Since the top of the interval was deeper than the existing Cari Covell 35 liner, a new liner was cemented to 3300 feet depth. The lower portion of the hole was sanded back to 3800 feet and plugged with cement to 3531 feet to contain the treatment in the desired interval. The frac string was to isolate the liner laps in the well from the treating pressure [49]. For well completion, a pre-perforated liner was installed in the treatment interval.

The hydraulic fracture treatment schedule is shown in Table3. Due to the high temperature of the geothermal system, special treatment design and materials were required for the stimulation. From a total of 7641 bbl frac ﬂuid, half was dedicated to wellbore and fracture pre-cooling; pumped at high rates in order to minimize its degradation [49].

Table 3: Well Baca 23 hydraulic stimulation treatment schedule [49].

Planned size Actual size Proppant Stage no. [bbl] [bbl] [lb/gal] [size] Fluid 1 4000 3582 8 - Water with ﬂuid loss additive (FLA) 2 500 502 8 - Polymer gel with FLA 3 500 502 2 100-mesh Polymer gel with FLA 4 500 526 6 - Polymer gel with FLA 5 900 905 1 20/40-mesh Polymer gel 6 1000 1000 2 20/40-mesh Polymer gel 7 600 562 3 20/40-mesh Polymer gel 8 58 62 6 - Water 8058 7641

A long term flow test showed a static temperature profile with a low bottom-hole temperature of 401°F (205°C), while a separate temperature survey showed a maximum temperature of 344°F (173°C) [49]. Therefore, two-phase flow was occurring and the temperature drop was associated with flashing in the formation [49]. In addition, a productivity test showed well recovery after each shut-in period followed by a decrease in mass flow and wellhead pressure. This concludes that permeability reduction associated with two-phase flow effects is most likely occurring due to partial closing of the fracture [49]. Microseis- mics were also monitored where a single fracture 100 m high and 160 m long might have been created, which is considered success as this fulfills the initial goals of the stimulation [48]. However it is concluded that the well is present in an impermeable formation and is unsuccessful in producing fluids for power generation [48]. 36 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

3.1.3.1.2 Well Baca 20 Shortly after the stimulation in well Baca 23, well Baca 20 was hydraulically stimulated. The hydraulic stimulation interval in the well was from 4880-5120 feet and the total depth of the well is 5827 feet with the packer set at 2412 feet [49]. At a temperature of 540°F (282°C), this was the hottest well to be stimulated in the DOE GRWSP [48]. The eleven stage treatment schedule was also similar to well Baca 23 with some minor adjustments. The proppant only consisted of sintered bauxite in equal quantities of 16/20-mesh and 12/20-mesh type. To stop leakage into the small natural fractures, 100-mesh calcium carbonate was pumped in stages 2-5 [49].

Temperature surveys showed a zone cooled by the natural frac fluids less than 100 feet in height near the bottom of the open interval of the well [49]. There was also a zone around 4720 feet depth that indicated some cooling most likely due to the workover fluids (brine) injected in stage 1, but could also be due to the frac fluids present in the lower part of the open interval [49]. Pressure build-up data indicated that a highly conductive fracture with a length of over 100 m was created, but productivity was poor and did not meet the needs for a commercial well [48].

3.1.3.2 Latera, Italy

The Latera geothermal ﬁeld is a water dominated hydrothermal system with temperatures of 210-230°C and was hydraulically and chemically stimulated in late 1983 as part of a geothermal well stimulation program set up by ENEL [50]. Hydraulic stimulation was performed on wells L1 and L6 as they proved to be dry after drilling [46][50].

Well L1 experienced several injection tests on the entire open hole section. The first three tests involved injection at rates between 45-97 m3/h but results indicated a relatively small fracture area [50]. Subsequent injection tests with increasing flow rates of 110, 220, and 340 m3/h were conducted for short intervals over 1-4 hours, and opened only about half the fracture area created by previous injection [50]. Finally a massive injection test at a rate of 310 m3/h over 17.5 hours increased the fracture area by a factor of 4 [50]. The main fracture did not completely close after injection, so short-term injectivity (in days) was improved [50]. However, a long term (1 month) injection test indicated no increase in fracture area [50]. The reason could be due to lack of connection between artificial fractures and the natural fracture network of the geothermal reservoir [46]. Clearly the main fractures in the well had potential to remain open for longer periods of time, but would have required inconceivable amounts of water to be injected. In addition, the transmissivity remained low but the skin factor went from positive to negative. No explanation is provided for the discrepancy of the results or whether they were measured after short- Cari Covell 37 term or long-term tests. Well L6 was stimulated in a similar matter as well L1, but the stimulation continued to require pressures too high for industrial injection rates and the operation was terminated early [50].

3.1.3.3 Salak, Indonesia

The Salak high temperature geothermal ﬁeld of Indonesia (240-315°C) has several wells that have experienced either hydraulic, thermal, or chemical (acid) stimulation. Only hydraulic stimulation is described in order to evaluate successes and failures of particular wells. Hydraulic stimulation was performed from 2007-2009 in wells Awi 18-1 and Awi 20-1 as they had low permeability west of the Cianten Caldera production area [51]. In addition, one of the ﬁrst zonal isolation modeling experiments took place from 2012-2013 in well Awi-3 as there was low injectivity after drilling [47].

Through injection of a large volume of water at high pressure, well Awi 18-1 showed decreasing wellhead pressure. The observation indicates conductivity improvements of existing fractures and/or newly developed fractures around the wellbore [51]. A phase of reinjection followed using condensate from the power plant. Although not directly considered a hydraulic stimulation, the reinjection had an effect on the 180% increase in injectivity index of the well. This suggests that not only was permeability improved, but a connection between the well and a low pressure system also occurred [51].

Well Awi 20-1 also experienced a decrease in wellhead pressure by 370 psi after the second injection stimulation period [51]. The decrease in wellhead pressure also corre- sponded to an increase in injectivity index as well Awi 18-1. However a ﬁnal injectivity test at the end of stimulation operations showed no improvement in injectivity index. This indicated that improvement in injection performance was mostly attributed to the estab- lishment of connectivity between the well and a lower pressure fracture network, rather than permeability improvement of existing fractures [51].

For well Awi-3, a conceptual model was built to study the beneﬁts of zonal isolation to improve well injectivity, then the actual stimulation was performed to compare results. The well was designed with a combined pre-perforated and blank 10-3/4” and 8-5/8” liner. The blank part was then cemented about 300 ft high to create zonal isolation. Further- more, the main injection casing 13-3/8” was successfully installed in one string, which eliminated tie-back (common in geothermal well), improving its long term reliability [47]. This combination of long string injection casing and zonal isolated liner would then allow the model to simulate injection at speciﬁc zones. The effects of injection were modeled in the lower zone and in the upper zone of the well (details regarding the location of each 38 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use zone are unavailable). The stimulation in the lower zone experienced some plugging and therefore no fracturing happened [47]. However, the upper zone stimulation improved early during injection and showed that the isolation method worked as intended [47]. Therefore the conceptual model for zonal isolation does not guarantee an improvement in injectivity, as the results are different in the lower and upper zones. During actual stimulation in the upper zone, initial injection capacity was around 150 kph at 500 psi wellhead pressure (WHP). After the stimulation, the well injectivity was tested to be 600 kph at 500 psi WHP. This huge improvement shows that large injection plays an important role in stimulating the reservoir, and has helped to eliminate the necessity of drilling another injection well [47]. However more work needs to be done to understand the vast difference between the simulation results of the lower and upper zone.

3.1.3.4 Mt. Apo, Philippines

Mt. Apo and other geothermal ﬁelds in the Philippines have experienced both hydraulic and acid stimulation, but only hydraulic stimulation will be discussed in this thesis. The geothermal ﬁeld averages reservoir temperatures greater than 300°C [52].

Hydraulic stimulation occurred in well K6 in November 2012 with the goal to increase well permeability [53]. At the end of drilling completion, the injectivity and transmissivity were low and high positive wellhead pressure was observed in addition to the low permeability of the well. River water and power plant condensates were injected continuously at 6 BPM and a controlled maximum pressure of 9.0 MPag over a period of one and a half months[53]. Based on injection ﬂow rate and wellhead pressure monitoring, no indication of permeability enhancement was achieved and no increase in injectivity index was observed [53].

Another well was hydraulically stimulated over ﬁve days from September-October 2013, but the literature does not indicate which well in particular. A decreasing trend in wellhead pressure was observed directly with increasing pump rates, signifying an enhancement of well permeability [54][55]. This was further supported by observing the changes in temperature gradient, where a decrease in temperature indicated a zone of cold water inﬁltration. The permeable zones noted before hydraulic stimulation were located between 2000-2050 mMD and at the bottom of the well [55]. After hydraulic stimulation, additional permeable zones were found between 950-1000 mMD, 1075-1150 mMD, and 1875-1950 mMD [55]. The injectivity index increased from 11.27 L/s-MPag to 14.34 L/s-MPag which also infers improvement in permeability [55]. Cari Covell 39

3.1.4 Enhanced geothermal systems (EGS)

The definition of EGS has been interpreted in several ways over time regarding rock types, depth, temperature, reservoir permeability, or type of stimulation technique involved. For the purposes of this thesis, the best definition is from the Bundesmin für Umwelt (BMU - The Federal Ministry for the Environment, Germany). The BMU defines enhanced geothermal systems as creating or enhancing a heat exchanger in deep and low permeable rocks of temperatures less than 200°C using stimulation methods [5]. Following BMU’s definition, Breede [56] further defines EGS as embracing hot dry rock, deep heat mining, hot wet rock, hot fractured rock, stimulated geothermal systems, and stimulated hydrothermal systems.

Stimulation methods that have been applied in EGS developments are summarized in Figure7, which shows that hydraulic stimulation is the most common method independent of rock type. Due to the relatively few cases where chemical or thermal stimulation technologies are applied, the definition of EGS is interpreted as only applying to hydraulically fractured systems [56]. Figure8 further shows the rock type and well depth of all the worldwide studied EGS projects as of 2013. According to both Figures7 and8, all fifteen igneous rock EGS fields have experienced hydraulic stimulation. However with further examination, the Bouillante EGS field of Guadeloupe has experienced thermal cracking; not hydraulic stimulation by definition of this thesis. The five sedimentary fields that have experienced hydraulic stimulation or have hydraulic stimulation programs planned are Mauerstetten, St. Gallen, Genesys Hannover, Genesys Horstberg, and Altheim. The two EGS fields of mixed igneous and sedimentary composition that have experienced hydraulic stimulation are Groß Schönebeck and Paralana. In addition, the metamorphic EGS field Bad Urach and the mixed igneous and metamorphic EGS field Desert Peak have both experienced hydraulic stimulation. A discrepancy is noted in the metamorphic category, where Larderello is stated as a thermal stimulation site. Larderello has actually experienced multiple acid stimulations since the 1980s, some of which are also based on the injection of water and could be interpreted as hydraulic stimulation [57]. These hydraulic stimulations are interpreted as being a part of a larger acid stimulation, therefore the Larderello field will not be discussed in this thesis. The Altheim site in Austria was too small of a project to have any significant record of stimulation. Lastly, the Raft River site in the United States will also be included in this discussion as preliminary results of hydraulic stimulation have been available for review as of 2014.

Each EGS field, with substantial literature to review, that has experienced hydraulic stimulation as an independent method is discussed based on the specific parameters used either in research or commercial applications, as seen in AppendixC. Essentially, each case is 40 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use unique and no conclusion is made about specific methods or procedures as suitable options for particular areas. Every project and location has characteristics that, at this time, are difficult to understand for application in low-temperature geothermal areas.

Figure 7: Stimulation methods applied to EGS projects worldwide as of 2013 [56].

Figure 8: Rock type and well depth of EGS projects worldwide as of 2013 [56]. Cari Covell 41

3.2 History of stimulation fracture modeling

Fracture modeling is important to determine the success (or failure) of a stimulation operation and is usually performed before stimulation commences. Two known cases in Germany and Canada are pointed out for fracture modeling using FRACPRO and MShale software. FRACPRO is typically used to address the prediction of pressure response in the well for planned stimulation treatments and the selection of appropriate equipment to handle expected wellhead pressures, friction, and near wellbore tortuosity [58]. Well GrSk4/05 at the Groß Schönebeck EGS site in Germany was first modeled in FRACPRO, then used as a real-time simulation tool. While fracture propagation was the goal of stimulation, the main purpose of FRACPRO was to model a treatment schedule [58]. In the Ft. McMurray area of Canada, research was performed with MShale to extract hot water from natural oil sands. A sensitivity analysis of estimating reservoir properties using prior knowledge about fracture location was done in order to determine the potential for an HDR project, and subsequently to determine optimal areas for stimulation [59]. Both cases were primarily concerned with modeling the effects of stimulation on fracture geometry. However, these software do not account for any potential production flow improvements after stimulation. A number of other fracture modeling software is available [60], but most model seismic activity with the goal to minimize environmental impact. Historically, the importance of fracture modeling was about the emphasis in fracturing low-permeability reservoirs in order to model the productive fracture length and the dimensionless fracture conductivity; i.e. the ratio of the ability of the fracture to carry flow divided by the ability of the formation to feed the fracture [7]. There is a need to de- rive a solution methodology from the oil and gas industry in order to take the effects of stimulation on fracture geometry and apply them to model production potential. This is the motivation for modeling in the MFrac suite for direct use purposes using geothermal resources, discussed further in the next chapter. 42 43

Chapter 4

Methods

This chapter will discuss the methods used regarding a case study for well HF-1 in Hoffell, Iceland. A lumpﬁt parameter model (LPM) was analyzed in order to measure the initial production potential of the well. This was done once before by Shengtao [61], but only across 5 months of data out of a 13 month long production test. The LPM compared the two data sets in order to more accurately provide a frame of reference for conducting a hydraulic stimulation model.

After determining that well HF-1 should be stimulated, a fracture model is done in MFrac software. This is to see the fracture effects after stimulation using two scenarios; via injection below a packer and above a packer. A third scenario of an open-hole was briefly considered, but ultimately rejected as the MFrac software cannot accommodate a stimulation interval as the same or similar size of the drillhole. The "frac fluid" used is a combination of a low concentration viscous gel to essentially mimic water, and a large grain size proppant of sand. The model also includes parameters for the treatment schedule, as well as initial conditions for modeling fluid loss and heat transfer effects.

MProd is then used to model the production ﬂow rate after the stimulation has been performed. This is done by using certain outputs from MFrac as inputs for MProd, along with known characteristics of the reservoir where the well is located. The MProd solution is based on a boundary condition for the pump rate used and net pressure measured after stimulation. An improvement ratio is then calibrated, where another LPM was modeled for a 10 year lifetime to see the well production improvement after stimulation. 44 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

4.1 Case study: Hoffell well HF-1

Iceland is a geologically young country (< 16 Myrs, [19]) lying on the Mid-Atlantic Ridge, which is the boundary between the North American and Eurasian tectonic plates. As a result of its location, Iceland is tectonically and volcanically very active with abun- dant geothermal resources associated with this activity. A map of the country is presented in Figure9, which shows the volcanic zone along the Mid-Atlantic Ridge and subsequently the geothermal areas classiﬁed into high temperature and low temperature areas. By deﬁnition, low temperature areas have a reservoir temperature below 150°C and high temperature areas have a reservoir temperature above 200°C [19]. Stimulation operations are commonly an integral part of the completion programs of geothermal wells drilled in Iceland, and are usually conducted at the end of drilling in order to enhance the output of the wells [19][62].

Figure 9: Locations of geothermal areas in Iceland based on reservoir temperature and geology [19].

The Hoffell case study area is located in SE-Iceland about 15 km outside the city of Hofn (pop. 1700), as shown in Figure 10. Hoffell is a low temperature geothermal field about 400 km east of Reykjavik, the capital city of Iceland, and is located at 64°42´20´Ń – 64°44´20´Ń and 15°04´20´É – 15°06´30´É. Within the region the extinct central volcano of Geitafell is found, but was active five million years ago [63]. Geothermal exploration in Hoffell began in 1992 with research done on surface geology, magnetic measurements, chemical analysis of the water, and geothermal gradient drilling [64]. The results Cari Covell 45 showed that there is potential of exploitable low temperature geothermal resources, as temperature gradients of up to 186°C/km were observed and chemical composition of the water indicated a 70-80°C temperature deep in the water system [64][65].

Figure 10: Map showing the location of the Hoffell case study area [63].

RARIK (Iceland State Electricity) drilled Well HF-1, but before the well was drilled in 2012 there were already 33 boreholes in the area with a cumulative total drilling depth of 3,594 meters [66]. The drilling of well HF-1 at Hoffell started in early November 2012 and lasted until January 11, 2013. The hole was first drilled down to 1,208 m depth, but was later deepened in February 2013, first to 1,404 m and finally to 1,608 m depth [66]. Figure 11 shows the location of well HF-1 with respect to the other boreholes in the Hoffell geothermal area. Most of the exploration wells were drilled N-S as surface manifestations indicated the main fault line to be in this direction. However when well HF-1 was drilled and tested, the free flow rate was very low at about 7 l/s. Recently, data loggers placed in the exploration boreholes on the east side of the geothermal field indicated that the main fault line is most likely oriented NE-SW, which may explain the low flow of well HF-1 [66]. 46 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Figure 11: Location of well HF-1 and some exploration wells [64].

After drilling was completed, long term production testing was performed to understand the reservoir behavior and to estimate its production potential [61]. The test started April 9, 2013. Water-level drawdown, production ﬂow rate and temperature were monitored and recorded continuously. The test concluded 13 months later on May 9, 2014. Data collected during this time was used as the basis for developing this thesis.

4.2 Lumpﬁt parameter model (LPM)

Lumped parameter modeling (LPM) is a simplified form of numerically modeling the hydrological properties of a low temperature geothermal reservoir. The observed changes in reservoir pressure (or water level) and the fluid production/injection rates can be matched using lumped parameter models, and consequently the fluid and/or energy production potential of a given field can be predicted [67]. For the software, Lumpfit beta was used, which was programmed by and obtained from Iceland GeoSurvey (ISOR) in 2015. The software is an update from PyLumpfit, programmed by ISOR in 2014. Cari Covell 47

4.2.1 LPM solution methodology

The theoretical foundation of lumped parameter modeling is the concept of dividing the geothermal system into two or more individual data blocks. LPM contrasts other numerical modeling softwares that use a range of 100-106 modeling blocks. Typically one block represents the production zone of a system, while the other blocks represent recharge zones. Within each block the properties are uniform, hence no internal differences exist. While assuming uniform tanks might seem disadvantageous in terms of simulation accuracy, LPMs actually provide a good result for modeling low temperature geothermal reservoirs [68]. No major drawbacks are seen from assuming uniformity because low temperature geothermal reservoirs are considered nearly isothermal and have almost uniform ﬂuid chemistry [67]. Lumped parameter modeling has been proven to quite accurately simulate various low temperature systems in Iceland [67][69][70].

An illustration of the representative tanks in LPM is shown in Figure 12, where a few tanks (capacitors) are connected by flow resistors (conductors). The tanks simulate the storage capacity of different parts of the reservoir and the resistors simulate the permeability. A tank in a lumped model has a storage coefficient κ when it responds to a load of liquid mass m with a pressure increase p = m/κ. The mass conductance (inverse of resistance) of a resistor is σ when it transfers Q = σ∆p units of liquid mass per unit time at the pressure difference ∆p. Withdrawal from the production tank will influence all connected tanks, according to set properties for individual tanks and connectors [69].

Figure 12: A general lumped parameter model used to simulate water level or pressure changes in a geothermal system. The three tank scenario is shown here [70].

Lumped models can either be open or closed. Open models are connected by a resistor to an inﬁnitely large imaginary reservoir, which maintains a constant pressure. When closed, lumped models are isolated from any external reservoirs. Actual reservoirs can most generally be represented and simulated by two- or three-tank closed or open lumped parameter models [69]. 48 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

The pressure response ∆p(t) of a single-tank open model for production Q, assuming a step response since time = 0, is given by the following equation [71]: Q σ1ti ∆p(t) = − 1 − e κ1 (6) σ1

The pressure response ∆p(t) of a more general open model with N tanks to a more constant production Q from time = 0 is given by:

N X Aj ∆p(t) = − Q 1 − e−Lj t (7) L j=1 j

The pressure response of a general closed model with N tanks is given by:

N−1 X Aj ∆p(t) = − Q 1 − e−Lj t − QBt (8) L j=1 j

Coefficients Aj, Lj and B are functions of the storage coefficients of the tanks (κj) and the conductance coefficients of resistors (σj) of the model, and can be estimated by the LUMPFIT program which uses an iterative non-linear inversion technique to fit a corresponding solution to the observed pressure or water level [69].

4.2.2 Initial production modeling of Hoffell HF-1

While the 13 month production test was being conducted, LPM work by Shengtao (United Nations University) was done for the first five months of production testing from April 9, 2013 to September 8, 2013 [61]. The LPM conducted in this thesis is based on data from the 13 month production test, but starting one month later from May 9, 2013 as data for the first month was deemed invalid. Pre-stimulation LPM results are then compared to the Shengtao analysis. Data was provided by ISOR with permission to use from RARIK. The purpose of performing a pre-stimulation production analysis with the additional production test data is to foresee a more accurate prediction of the required flow for sustaining production over a certain period of time. The required flow calculated in Lumpfit beta will be used as a guideline for determining the effects of the post-stimulation treatment.

In Shengtao’s analysis, two lumped parameter models, a two-tank closed model and a two-tank open model, were used to simulate the five month production data from well HF-1. During the 152 day production process, production started with a flow rate of 20 Cari Covell 49 l/s, and was later changed to 15 l/s by August 2, 2013. The water level in the well varied from -80 m to -140 m depth (below surface level). An average reservoir temperature of 72°C was assumed based on the measured data [61]. The two models were chosen as they provided a good fit between the measured and calculated water level in the well, which can be seen in Figure 13. In the analysis for this thesis, only the two-tank open model was considered, as the two-tank closed model did not provide a good fit of the measured and calculated water level in the well, as shown in Figure 14. This may be a result of the uncertainty of LPM within the Lumpfit beta software [68]. In addition to the five month data, production continued with a change in flow rate to 5 l/s by November 13, 2013, 3 l/s by December 3, 2013, and 1 l/s by April 30, 2014 as seen in Figure 15. The water level in the well varied from -5 m to -140 m depth (below the surface level). An average reservoir temperature of 69°C was assumed based on the measured data.

Figure 13: Monitored and calculated water level of Well HF-1 from April 9, 2013 to September 8, 2013 of the long-term production test. Calculated values are those of the LPM, where the left shows the two-tank closed model and the right shows the two-tank open model. Time t = 0 corresponds to April 9, 2013 [61]. 50 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Figure 14: Monitored and calculated water level of Well HF-1 from May 9, 2013 to May 8, 2014 of the long-term production test. Calculated values are those of the LPM, where the left shows the two-tank closed model and the right shows the two-tank open model. Time t = 0 corresponds to May 9, 2013.

Figure 15: Long-term production test for a one year period of well HF-1. Time t = 0 corresponds to May 9, 2013.

Based on the lumped parameter models established above, future predictions could be calculated to estimate the response of the water level (reservoir pressure) to exploitation. Ten year predictions were calculated to help gain an understanding of the general water level changes for different production ﬂow rates [61]. A Monte Carlo simulation performed by Shengtao shows that if the geothermal heating system will be used for 50 years with average thermal power of 6.0 MWth, a thermal water ﬂow rate of 28.6 l/s is needed; and if the geothermal heating system will be used for 100 years with a thermal power of 3.0

MWth, a thermal water ﬂow rate of 14.3 l/s is needed [61]. The Monte Carlo simulation results form the basis for the prediction production scenarios; therefore the production rates were set to 28.6 l/s, 21.4 l/s, 14.3 l/s and 7.15 l/s [61]. Cari Covell 51

As seen in Figure 16, the set of data over the ﬁve month period shows that the water level behaves quite differently in the two models. Over the next 10 years, the water level is predicted to decline very sharply in the closed system while in the open system it reaches equilibrium. As seen in Figure 17, the year-long analysis for this thesis also shows the open system as reaching equilibrium.

Figure 16: Predicted water levels in well HF-1 for the next 10 years for different production rates using the ﬁve month period long-term production test data. Conservative predictions using two-tank closed model are on the left. Optimistic predictions using two-tank open model are on the right [61].

Figure 17: Predicted water levels in well HF-1 for the next 10 years for different production rates using the year-long period of long-term production test data. The optimistic two-tank open model is shown.

As seen in Table4, the water level behaves quite differently between the closed system and the open system. With an increasing production rate, the difference between the closed tank system and the open tank system becomes more obvious [61]. The water level in a closed system had a greater response to large production rates than that in an open 52 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use system. With the pump at a depth of 175 m, Shengtao’s LPM model shows the closed tank system requiring a sustained production rate of 6.7 l/s and the open tank system requiring a sustained production rate of 26.8 l/s. The LPM model used in this thesis shows the open tank system requiring a sustained production rate of about 28.6 l/s with the pump at a depth of 175 m.

Table 4: Predicted water levels in well HF-1 after 10 years production [m]

Based on ﬁve months data [61] Based on year-long data Production ﬂow rate Conservative model Optimistic model Optimistic model [L/s] (closed system) (open system) (open system) 7.15 -187 -6.5 -44 14.3 -421 -81.6 -88 21.45 -639 -140 -132 28.6 -841 -183 -176

A production rate as high as 28.6 l/s would lead to a very great water level decline if the geothermal system was a closed system; and a rate this large would cause the water level to drop down to -841 m after 10 years, which is not realistic [61]. However, it is also very unlikely that the system is completely closed. Furthermore, the difference in production rate for each of the optimistic approaches is very small (about 2 l/s), so it is shown that with more production data the results of the 10 year prediction are relatively the same in this case. Although the conservative approach was unable to be modeled with the year- long production data, it can be assumed that the production rate would be relatively close if a better ﬁt were to be made. From these results, the behavior of well HF-1 is most likely to be between the 15-20 l/s production range. However, to reduce negative inﬂuence, reinjection or stimulation will be necessary, especially for large production rates if the system turns out to be relatively closed [61]. This is the underlying reason for creating a stimulation treatment program for well HF-1.

4.3 MFrac model

MFrac Suite Hydraulic Fracturing Software is a comprehensive design and evaluation simulator containing a variety of options including three-dimensional fracture geometry, auto design features, and integrated acid fracturing solutions; originally designed for the oil and gas industry. Fully coupled proppant transport and heat transfer routines per- mit use of the program for fracture design, as well as treatment analysis [60]. MFrac is Cari Covell 53 not a fully 3-D model, but rather is formulated between a pseudo-3D and full 3-D type model with an applicable half-length to half-height aspect ratio greater than about 1/3 [60]. MFrac also has options for 2-D type fracture models in the form of Geertsma & de Klerk (GDK) and Perkins & Kern (PKN). MFrac is the calculation engine for real-time and replay fracture simulation and works in conjunction with the real-time data acquisi- tion and display program MView. The MFrac suite also includes MPwri, MinFrac, MFast, MProd, MNpv, MLite, MWell, MShale, MACQ, and MDBE. For the purposes of this thesis, only MProd was used as an additional software for analysis to MFrac, discussed further in section 4.4. The MFrac suite was obtained from Baker Hughes Incorporated with a full academic license, and is programmed by several third parties [72].

4.3.1 MFrac governing equations

The fracture propagation solution is obtained numerically by satisfying mass conservation, continuity, momentum, width-opening pressure elasticity condition, and the fracture propagation criteria; where a detailed description of these equations and the solution methodology is provided by Meyer et. al [60][73][74][75][76]. All nomenclature is listed in the preamble of this thesis.

4.3.1.1 Mass conservation

The governing mass conservation equation for an incompressible slurry in a fracture is deﬁned by: Z t q(τ)dτ − Vf (t) − Vl(t) − Vsp(t) = 0 (9) 0 where

Z t Z A C(A, t) Vl(t) = 2 dAdt αc 0 0 [t − τ(A)]

Vsp(t) = 2SpA(t) τ(A) = t[A/A(t)]αa .

The above mass conservation equations are solved numerically in MFrac by elementally descritizing a fracture grid and then integrating over each element. The above equations for performing minifrac analysis (i.e. from a previous fracture) can be simpliﬁed for 2-D 54 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use type models for ﬂuid loss due to leakoff during and after pumping:

During Pumping √ Vl(t) = πC(t)A(t) tΦ(αaαc)

After Pumping p Vl(θ) = 2C(tp)A(tp) tpG(αaαc, θ)

where θ = t/tp.

4.3.1.2 Mass continuity

The mass continuity equation in terms of the ﬂow rate per unit length q = vW is:

∂W ∇~ · ~q + 2q + = 0 (10) L ∂t

~ where ∇ ·~q = ∂qL/∂x + ∂z/∂z and ∂L is the leakoff rate per unit leakoff area (i.e. leakoff velocity).

4.3.1.3 Momentum conservation

The momentum equation (equation of motion) for steady ﬂow is:

∇~ P = −(1/2)fρ~q2/w3 (11) where

f = 24/Re ; laminar ﬂow, f = fR(e, ) ; turbulent ﬂow,

and f is further deﬁned as the Darcy friction factor, Re as the Reynolds number, and as the relative wall roughness. Cari Covell 55

4.3.1.4 Width-opening pressure elasticity condition

The crack-opening and opening pressure relationship is of the form:

2(1 − v) W (x, z, t) = Γ (x, y, z, t) H ∆P (x, 0, t) (12) W G ζ where ΓW is a generalized inﬂuence function, Hζ is a characteristic half-height, and ∆P is the net fracture pressure P − σ.

4.3.1.5 Fracture propagation criteria

The criteria for fracture propagation is based on the concept of a stress intensity factor

κI . The fracture will propagate when the stress intensity factor equals the fracture toughness κIC or critical stress intensity of the rock (κI = κIC or σI = σIC , whichever is greater).

4.3.2 MFrac solution methodology

The governing differential equations for fracture propagation are differentiated with respect to time and then simpliﬁed by the transformations:

t dL(t) t dA(t) α ≡ ; α ≡ L L(t) dt a A(t) dt

t dWw(t) t dHw(t) αw ≡ ; αH ≡ Ww(t) dt Hw(t) dt t d∆P (t) t dC(t) α ≡ ; α ≡ P ∆P (t) dt c C(t) dt to form a set of equations in terms of the alpha parameters (αζ = t/ζdζ/dt). The length propagation parameter is of the form:

1 − (αca + 1/2)(1 − η) + αterm αL = 0 (13) η(1+βH (3+n )) 1 + 0 (n +1)(1−βγ ) where αζ accounts for the time dependent gamma parameters, non-steady injection rates and ﬂuid rheology, spurt loss, fracture toughness, etc. The fracture efﬁciency is given by η and βH = αH /αL. The geometric factor βγ is equal to unity for the PKN and 3-D type fracture models and equal to zero for the GDK model. Additional alpha parameters 56 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use for 2-D type fractures are also given by [60]. The previous equation and the formulated constitutive relationships control the time dependent length propagation solution:

αL(t) L(t) = L(tn)(t/tn)) . (14)

4.3.3 Stimulation set-up

For Hoffell well HF-1, a 3-D fracture simulation model along with a complete proppant transport and ﬂuid treatment schedule were constructed in MFrac. The goal was to create a stimulation program based on the characteristics of the well and the geothermal area in order to analyze the effects of fracture propagation. The target area was a fracture located at 1093 m depth as televiewer logs indicate good cracking and is therefore the best candidate for stimulation [77]. Therefore, it was decided to place the packer in a conservative interval of 1070 m depth to 1110 m depth; allowing for a 40 m range of placement. Injection down the casing will be done through a 12 hour stimulation in order to show one cycle of a typical stimulation program. Proppant type used was a 20/40 mesh Jordan sand and ﬂuid type was a low concentration water based gel, where each of their properties can be seen further in AppendixD. The idea of creating a water frac (WF) that has low concentrations of sand and gel will aid in creating a long fracture to intersect the main fault line. Two scenarios are created: 1) injection below a packer and 2) injection above a packer.

4.3.4 Governing model parameters

The first step in MFrac is to indicate the type of stimulation desired. This is designated in three stages: 1)general, 2) fracture, and 3) proppant. General conditions include the assumption of linear reservoir coupling as a one-dimensional analysis, as this option is the most common fluid loss model used for propagating fractures. Subsequently, fluid loss is to remain constant. The treatment design schedule is in auto design in order to optimize injection time. Wellbore hydraulics will be tabulated using an empirical model, where the wellbore friction factor is used to determine the energy dissipation (pressure loss) in the wellbore [73]. With this, three distinct types of behavior are possible with the combined correlation used in MFrac, and are summarized in the expressions for the Fanning friction factor in Table5. Through an iterative process, MFrac determines which correlation is most applicable in determining the friction factor. The criteria are based on the argument √ that 1/ f will always be greater than the Prandtl-Karman Law (lower bound) and less Cari Covell 57 than Virk’s maximum drag reduction asymptote (upper bound). Therefore, when the transitional correlation developed by Keck et al. [78] reaches either the upper or lower bound, it is automatically adjusted to meet the above criteria, as seen in Figure 18.

Table 5: Fanning friction factors √ √1 Maximum Drag Reduction [79] f = 19log(Res f) − 32.4 1 √ Transitional ﬂow [78] √ = Alog(Res f + B f √ √1 No Drag Reduction [80] f = 4log(Res f) − 0.4

Figure 18: MFrac Pipe Friction Empirical Correlations [72].

For the fracture dialog box, no flowback is assumed as no negative flow rate is present. The model will not simulate to closure after pumping, therefore it is assumed that the fracture remains open after treatment. The fracture fluid gradient option is turned on to tabulate hydrostatic pressure changes as a function of depth. Default growth of the fracture is assumed (i.e. in the positive and negative direction) as this will measure pressure decline even after/during pump shut down. This tends to happen in stimulation treatments where not enough fluid is available on site and re-filling of tanks is necessary. The fracture initiation interval will be calculated using the minimum stress interval, which is 10% of the initial fracture height [72]. This option calculates effective closure pressure to keep the fracture open over the interval. Creating a fracture friction model will not be necessary as laminar flow normally exists in the fracture. For this case, the classical solution for fluid flow is used and the Darcy friction factor is f D = 24/Re [72]. The inclusion 58 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use of wall roughness is turned off, as the Darcy friction factor inside the fracture is used without modification. This selection assumes that the fracture surface is a smooth planar feature without roughness [72]. Tip effects will not be analyzed, as a constant pump rate was used and it can be assumed that there is no net pressure at the fracture tip (see section 2.1.1.1.3 for further information on this theory).

For the proppant dialog box, the proppant ramp (i.e. changes in proppant concentration as a function of time) is turned off because a uniform proppant concentration will be used throughout the stimulation. Perforation erosion will not be considered as the stimulation will model as close to an open hole as possible. The model will use a conventional (link proppant) transport methodology because this couples the proppant solution with fracture propagation and is the optimal solution of proppant transport when using the Auto Design treatment schedule feature. The proppant settling option will be user speciﬁed as the size of the particles used is typically much smaller than for oil and gas applications. Fracture- proppant effects control the effect of proppant concentration on frictional pressure losses in the fracture, and will use an empirical solution. The procedure involves a correction to the base viscosity to produce a relative viscosity term. Once this is done, the friction factor is calculated based on the fracture friction model selected (in this case, the classical Darcy friction factor solution). The general correlation as a function of proppant void fraction is:

αµ µr = (1 − φ) where

µr = relative slurry velocity

αµ = exponent coefﬁcient φ = proppant void or particle volume fraction

4.3.5 Wellbore hydraulics

The wellbore hydraulics are deﬁned to essentially build the well in order to calculate its total volume. First, the casing depths and dimensions are set for Hoffell well HF-1 based on drilling reports as seen in Table6[66]. The outside diameter (OD) is speciﬁed using the MFrac internal database, where the subsequent weight and inside diameter (ID) are calculated. The lightest weight for each OD was then chosen. Although the wellbore is actually open hole after the production casing of 400 m depth, it is necessary to include the corresponding depths where additional drilling was performed in order to construct a Cari Covell 59 complete well in MFrac. Drilling depths to 1404 m and 1608 m were reached using a drill bit. Up to 1404 m depth a drill bit size of 9 7/8" was used, however casings are not made in this dimension and a 9 5/8" was assumed. A drill bit size of 8 5/8" was used for the remaining 202 m to 1608 m depth. The difference in measured length and section length are not applicable for this geothermal well, as the study does not account for the use of a liner where section length is different than measured depth.

Table 6: Casing Dimensions for Hoffell well HF-1

Measured depth Section length OD Weight ID [m] [m] [in] [lbf/ft] [in] 3.9 3.9 20 94 19.124 23.8 23.8 14 50 13.344 400 400 10.75 45.5 9.95 1404 1404 9.625 29.3 9.063 1608 1608 8.625 20 8.191

Next, the deviation of the well is built in MFrac. The wellbore survey method for determining deviation will use the average angle method, where well path between 2 stations is along a straight line; and whose length is the measured depth difference between the 2 stations and whose inclination and azimuth angles are the average of the station’s values. Since the ISOR drilling report only gives a graph representing horizontal deviation, the inclination angle was calculated in different iterations as a slope of the line of horizontal deviation versus measured depth [77]. All values were entered as measured depth (MD) and total vertical depth (TVD) was internally calculated in MFrac, as seen in Table7. Total deviation in the well is approximately 8°. Once the casing dimensions and well deviation are set, a cross section of the well is created as seen in Figure 19.

Table 7: Hoffell well HF-1 deviation

Measured depth Inclination angle TVD [m] [°] [m] 580 0.796 579.986 720 2.045 719.943 1080 1.989 1079.72 1200 2.386 1199.63 1280 3.576 1279.52 1380 5.711 1379.20 1600 7.765 1597.68 60 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Figure 19: Wellbore cross section for Hoffell well HF-1.

Bottom-hole treating pressure (BHTP) was speciﬁed by MD and 3 entries are allotted in MFrac. This will essentially create an output that measures BHTP from these 3 points in the well. The three points chosen were depth of production casing (400 m), depth of packer placement (1100 m), and the bottom of the well (1608 m).

Lastly, the total volume of the well is calculated based on wellbore configuration, surface line volume, and the wellbore volume reference depth using the governing equations for mass conservation. Surface line volume is the volume of fluid/slurry in the service line(s) upstream of the well entrance, and is assumed as zero because all treatment parameters are referenced at the well entrance. The wellbore volume reference was entered as MD and TVD is subsequently calculated, but this value has to be above the bottom of the casing and is therefore defined at approximately 1600 m depth.

4.3.6 Rock properties

Rock type is described in the drilling reports from ISOR [66]. For simplification purposes, rock type was defined in three layers based on MD, as in reality there are only slight differences in the properties of various rock types within the major governing layers of the well. The three layers of gravel, dolomite, and intrusive volcanics along with their Cari Covell 61 governing properties are shown in Table8. Stress and stress gradient were calculated based on internal calculations in MFrac (see [73]). Young’s modulus, poisson’s ratio, and fracture toughness vary in range for different rock types, where typical values can be seen in the Baker Hughes MFrac manual [72]. The critical stress, defined as the minimum critical stress for the fracture to propagate in the vicinity of a constant stress field, is set to zero via the database option; i.e. only fracture toughness will be considered. Critical stress may also be thought of as the apparent tensile strength since it is the critical stress that must be over come for the crack to propagate (in a uniform stress field) [72].

Table 8: Rock properties of Hoffell well HF-1

Zone TVD at MD at Stress Stress Young’s Poisson’s Fracture Critical name bottom bottom gradient modulus ratio toughness stress [m] [m] [psi/ft] [psi] [psi] [psi-in1/2] [psi] Gravel 17.999 18 0.6 35.4322 4.5e+06 0.2 1000 0 Dolomite 1099.71 1100 0.83 2994.6 7e+06 0.13 1000 0 Int. 1605.6 1608 0.7 3687.41 5e+06 0.25 1000 0 volcanic

4.3.7 Zones data

The zone of stimulation for each case below and above the packer must be speciﬁed. Typically a packer is placed where the fracture can be isolated within a 30-50 m interval. As the main fracture for well HF-1 is located at 1093 m depth, the packer would be isolated somewhere between 1070-1110 m depth. Therefore, stimulation below the packer is deﬁned within a zone that starts at 1110 m depth and goes to the bottom of the hole at 1608 m depth. The zone of stimulation above the packer starts at the end of the production casing at 400 m depth and goes to 1070 m depth.

Within the zones module of MFrac, perforation data is necessary to enter as MFrac caters to oil and gas wells. In essence, the number of perforations is set to 2000 at 0.75 in diameter in order to mimic an open hole geothermal well. The pay zone (length of stimulation interval) has an associated permeability for each case below and above the packer. The permeability used below the packer in the intrusive volcanic zone is low at around 10−4 darcy, while the permeability above the packer in the dolomite zone is higher at around 10−1 darcy [81]. 62 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

4.3.8 Treatment schedule

The total slurry volume injected was calculated assuming a constant 60 l/s flow for 12 hours and a total of 15 tons of proppant. The proppant distribution style was set using the maximum proppant concentration because a constant proppant concentration is assumed throughout the stimulation, therefore MFrac will design treatments with the last propped stage at the final proppant concentration specified. This option also creates a treatment schedule (when in auto design mode) where stages do not screen or bridge out and the maximum proppant concentration in the fracture will not exceed the maximum value specified. The proppant settling rate was obtained using Figure 20, assuming 0.8 mm diameter sphere/grain size [58], at 20°C injection fluid temperature. Table9 shows each of the properties defined in the treatment schedule.

Figure 20: Velocity of cuttings in mm/s [58].

Table 9: Treatment schedule for Hoffell well HF-1.

Property Value Unit Slurry volume 2606 m3 Pump rate 3600 L/min Initial and incremental proppant concentration 0.571429 lbm/gal Final proppant concentration 0.571429 lbm/gal Maximum proppant concentration (at tip) 0.571429 lbm/gal Proppant settling rate 10 cm/s Cari Covell 63

4.3.9 Fluid loss

To model fluid loss from the fracture into the reservoir and surrounding layers, additional information characterizing the formation and in-situ diffusivity parameters is necessary. The leakoff coefficient needs to be specified for each rock layer and the Baker Hughes MFrac manual provides a range of values for any given formation. Since the Hoffell geothermal area is dominated by dolomite and intrusive volcanics, a range of 0.001 to 0.003 ft/min1/2 is acceptable and a value of 0.002 ft/min1/2 was therefore assumed. Spurt loss are assumed to be zero as this is only present for wall-building fluids [7].

Fluid loss is also pressure dependent and is based off of Table 10. Pressure was calculated at zones of BHTP measurements located at 400 m, 1100 m, and 1608 m depth via production test data. Arbitrary leakoff multipliers are assumed for the three zones and gradually decrease as depth increases, where zones are cooled and gradually take longer to heat up. The leakoff coefﬁcient throughout the stimulation therefore changes and is calculated as a function of pressure. This is helpful for modeling leakoff in naturally fractured reservoirs. While fracturing a naturally fractured formation, the pressure in the fracture may approach the critical pressure. When the critical pressure of the formation is reached, natural fractures open and accelerated leakoff occurs.

Table 10: Pressure dependent ﬂuid loss for well HF-1.

Pressure Leakoff coef. Spurt loss coef. [bar] multiplier multiplier 57.5023 0.75 1 109.971 0.15 1 154.029 0.1 1

4.3.10 Proppant criteria

The minimum number of proppant layers to prevent bridging is the minimum number of layers in the fracture at which bridging occurs. In MFrac, a bridge-out is assumed to occur if the average fracture width integrated over the fracture height is less than the minimum number of proppant layers to prevent bridging. This makes sure that the fracture width is greater than the bridging criteria in order for the proppant to pass through. Typically, a value of 1.5 to 3 is used [72]. The minimum concentration/area for propped fracture means that anything below this concentration after embedment is included and the fracture will not be reported as being “propped”. A typical value ranges from 0 to 0.2 lbm/ft2 (1.0 64 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use kg/m2)[72]. The amount of embedment depends upon the proppant and formation type, and MFrac assumes this embedment is in the fracture at closure, therefore a value of zero is used. Finally, closure pressure on the proppant is calculated using the bottomhole ﬂuid pressure subtracted from the horizontal stress of the area. The horizontal stress of a geothermal area similar to Hoffell in Iceland was assumed as 500 bar [81]. The bottomhole ﬂuid pressure of 154 bar was taken from the production data of well HF-1. Therefore, the closure pressure on the proppant is approximately 344 bar. This is used to determine the proppant permeability which is interpolated from the proppant database. As the closure pressure (stress) increases, the proppant pack permeability decreases. Table 11 further illustrates the required proppant criteria.

Table 11: Proppant criteria for well HF-1.

Property Value Unit Min. number of proppant layers to prevent bridging 2 Min. concentration/area for propped fracture 0.1 [lbm/ft2] Embedment concentration/area 0 [lbm/ft2] Closure pressure on proppant 344 bar [psi]

4.3.11 Heat transfer

An analytical heat transfer model is included in MFrac that combines thermal convection in the fracture, with transient conduction and convection in the reservoir. MFrac predicts the heat-up of the fracturing fluid within the wellbore and/or the exchange of heat between the fluid and the reservoir during fracture propagation. MFrac couples the heat transfer, fluid flow, and fracture propagation expressions to characterize the time dependent fracture temperature profiles.

The fluid inlet is to be at the surface as MFrac will calculate heat transfer in the wellbore and in the fracture using this option. The reservoir lithology must be selected from the MFrac database, and the closest option to the Hoffell field is dolomite, where average porosity is subsequently applied within the MFrac program. The mean formation temperature of the area is 69°C as measured from production test data, and the injection fluid temperature is assumed to be 20°C. Table 12 further lists heat transfer properties. Cari Covell 65

Table 12: Heat transfer properties for Hoffell geothermal ﬁeld.

Property Value Unit Fluid inlet Surface Base fluid type Water Reservoir lithology Dolomite In-situ fluid type Water Average porosity 0.15 fraction Mean formation temperature 69 °C Injection fluid temperature 20 °C

4.4 MProd model

MProd is a single phase analytical production simulator. Although the program was designed primarily for hydraulic fracturing applications, it can also be used to explore the production potential of unfractured reservoirs. MProd has options for Production Simu- lation, History Match Production Simulation, and Fracture Design Optimization. Produc- tion Simulation will be used in this thesis which allows the user to input typical production data to simulate well performance for fractured and unfractured wells. The capability to compare the output (numerical simulated results) with measured data is also provided. MProd is integrated and fully compatible with MFrac to provide full feature optimization, where the output produced by MFrac can be used by MProd. The numerical results of MProd, in turn, can be imported by MNpv to perform economic analyses, but will not be done in this thesis.

4.4.1 MProd governing equations

The governing equations for simulating production from fractured and unfractured wells in closed and infinite reservoirs are based on the Baker Hughes manual solution methodology [72] presented in the following sections. The production solution is obtained numerically by satisfying conditions related to pre-defined dimensionless parameters, pseudopressure, the trilinear solution, pseudosteady-state pressure and resistivity solutions, wellbore choked skin effect, the pseudo-radial flow solution, productivity increase, and desuperposition. When appropriate, the solution methodology is also given. All nomenclature is listed in the preamble of this thesis. 66 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

4.4.1.1 Dimensionless parameters

The dimensionless pressure pD for a constant production rate q is deﬁned as:

2πkh p = ∆p (15) D qµ where k is the formation permeability, h is the formation height, µ is the reservoir viscosity, and ∆p = pi − pwf is the differential pressure (the initial reservoir pressure pi minus the ﬂowing pressure pwf ).

The dimensionless times based on the drainage area A, wellbore radius rw, and fracture half-length xf are deﬁned as:

kt kt kt t = t = t = DA , D 2 , and Dxf 2 ctφµA ctφµrw ctφµxf where φ is the formation porosity and ct is the formation compressibility.

The dimensionless rate qD for a constant flowing pressure pwf is defined as: µ q = ·q (16) D 2πkh∆p (t) and the flow rate as a function of the dimensionless rate is:

2πkh∆p q = ·q (17) (t) µ D where ∆p = pi − pwf is the constant drawdown pressure.

The productivity index J is deﬁned as:

q 2πkh J = = JD (18) p¯ − pwf µ where q is the ﬂow rate, p¯is the average reservoir pressure, and pwf is the ﬂowing pressure.

The dimensionless productivity index JD is deﬁned as: µ µ q JD = ·J = · (19) 2πkh 2πkh p¯ − pwf Cari Covell 67

The dimensionless and Laplace parameters used in the production model are:

kf wf C C CfD = , CD = 2 , CDf = 2 , kxf 2πφcthrw 2πφcthxf

C1 = η/ηf , η = k/(φµct), ηf = kf /(φµct)f , π ys k sf = − 1 , 2 xf ks 2 −π a = , b = . CfD CfD

The Laplacian operator used in the production models is given by s. The fracture skin is given by sf .

4.4.1.2 Pseudopressure

The real gas potential or pseudopressure is deﬁned as:

Z p p m(p) = 2 dp (20) pb µ(p)Z(p) where pb is an arbitrary base pressure and Z is the real gas deviation factor. The real gas pseudopressure equation can be simpliﬁed for certain pressure ranges. At low pressures µZ is essentially constant, while at higher pressures it is directly proportional to pressure [82].

At low pressure (p < 2000 psi), the real gas pseudopressure based on a constant µZ product is:

2 2 2 2 (p − pb ) (pi − pwf ) m(p) = or m(pi) = µZ µiZi where µi and Zi are initial condition gas properties at pi.

At high pressures (p > 3000 psi), the real gas pseudopressure based on a constant p/(µZ) product is:

2p(p − pb) 2pi(pi − pwf ) m(p) = or m(pi) = . µZ µiZi

When the ﬂuid type is gas and the Internal PVT correlations are not used, the model requires that a table of µ and Z be entered as a function of pressure. The pseudopressure function is then automatically used to calculate the dimensionless pressure. 68 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

4.4.1.3 Trilinear solution

The trilinear solution for a ﬁnite-conductivity fracture in Laplace space is for [83][84]:

Constant ﬂow rate b pD(s) = , or for s(sbCDf − (ΨtanhΨ) Constant pressure 1 Ψ qD(s) = 2 = − tanhΨ. s pD(s) sb

The Ψ parameter used in the above equations is deﬁned by:

s a(s + s1/2)1/2 Ψ = 1/2 1/2 + C1s (21) 1 + (s + s ) Sf

This analytical solution is based on transforming the equations above from Laplace space to real time using the Stehfest inversion algorithm [85].

4.4.1.4 Pseudosteady-state pressure and resistivity solutions

The pseudosteady-state dimensionless pressure solution in a closed system can be written as [86]:

pD = 2πtDA + 1/JD (22) where the inverse productivity index is given by

4A 1/J = 1/2ln . (23) D γ 0 2 e CArw

The pseudosteady-state resistivity solution for a ﬁnite-conductivity fracture in a closed system in terms of the inverse productivity index is [87]: 1 xe = ln βxe + f (24) JD xf where the pseudo-skin function f is

π f = ln + ζ∞ . (25) CfDg(λ) Cari Covell 69

The pseudosteady-state resistivity equations are solved to generate the production solution for a single fracture in a closed rectangular reservoir for all times (i.e., linear, bilinear, trilinear, and pseudosteady-state). These generated fundamental solutions for constant rate and constant pressure boundary conditions are then used to calculate the composite multiple transverse fracture solution.

4.4.1.5 Wellbore choked skin effect

Mukherjee and Economides [88] identiﬁed that the inadequate contact between a vertical transverse fracture and the horizontal well resulted in a restriction that can be quantiﬁed by a choked skin effect as given by:

kh h π sch = ln − (26) kf wf 2rw 2 in terms of the dimensionless fracture conductivity. This illustrates that as the height interval to wellbore radius ratio or height to propped length ratio decreases (i.e., radial to linear ﬂow in the fracture) or the dimensionless fracture conductivity increases, the skin due to convergence becomes smaller. Soliman et al. [89] concluded that a high conductivity tail-in could be incorporated to reduce the additional pressure drop because of ﬂow convergence around the wellbore.

4.4.1.6 Pseudo-radial ﬂow solution

The non-dimensional pressure drop at the wellbore, for an unfractured well, is [82]:

1 1 pd = − Ei − (27) 2 4tDw where the exponential integral E − i is deﬁned as

Z ∞ e−µ Ei(−x) = − dµ. (28) x µ

This solution is also used for the pseudo-radial solution of the fractured well by matching the pseudo-radial and trilinear solutions. 70 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

4.4.1.7 Productivity increase

The productivity increase as deﬁned by the ratio of the productivity indices for the fractured and unfractured wells as given by [73] below.

4.4.1.7.1 Constant ﬂow rate The instantaneous (current rate) and average (volume) productivity indices are:

Instantaneous

J/J0t = pDw /pDf Average Z Z

J/J0|V = pD0 dt/ pDf dt

4.4.1.7.2 Constant pressure The instantaneous (rate) and average (volume) productivity indices are:

Instantaneous

J/J0t = qDf /qD0 Average Z Z

J/J0|V = qDf dt/ qD0 dt where f refers to the fractured well and 0 to the unfractured case. For unfractured reservoirs the subscript f refers to the well with no skin. When considering a series of constant rate or pressure changes, the productivity parameters outlined above are equal to the equivalent values calculated by superposition.

4.4.1.8 Desuperposition

The concept of desuperposition has been illustrated by Gringarten, Ramey and Raghavan

[90] for modifying known values of pD to dimensionless pressures describing somewhat different systems. This method is used to calculate the effect of skin and fractures in closed systems. For closed systems, the dimensionless pressure pd is found from the following relationship:

pd(CD,S) = pD(CD = 0,S = 0) − pD∞(CD = 0,S = 0) − pD∞(CD,S). (29) Cari Covell 71

In the above equation pD is the dimensionless pressure for a closed system and pD∞ is the dimensionless pressure for an inﬁnite (unbounded) reservoir. The ﬁrst term on the right side of the equation is for the closed system with zero skin and zero wellbore storage.

Dimensionless pressure pD∞ for a single well in an inﬁnite system with zero skin and wellbore storage is subtracted from this dimensionless pressure term. This dimensionless pressure pD∞ for a single well in an inﬁnite system with the desired wellbore storage and skin factor is then added.

4.4.2 Stimulation set-up

The goal of MProd is to take the output from MFrac in order to simulate a production test performed after two scenarios of stimulation; below and above a packer. The solution is for the case of a single fracture in an infinite reservoir (i.e. open tank) where the production boundary condition is based on the net flowing pressure output of MFrac. The original production data based on flow rate from HF-1 is then overlayed for comparison in order to obtain a productivity index.

4.4.3 Formation data

The formation data module provides a location for entering the reservoir properties needed to perform a simulation. The total pay zone height is entered based on the scenario being analyzed, followed by all reservoir properties that were obtained via a Monte Carlo simulation performed by Shengtao [61]. Table 13 shows the input dialog box.

Table 13: Formation data for well HF-1.

Property Value Unit Total pay zone height 498 (below) or 670 (above) [m] Equivalent reservoir permeability 5.686 [mD] Initial reservoir pressure 38.5 [bar] Total reservoir compressibility 6.665e-08 [1/kPa] Equivalent reservoir porosity 10 % Equivalent reservoir viscosity 0.000399 Pa·s 72 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

4.4.4 Single case fracture characteristics

A calculated radial option is available for single case fracture characteristics in order to calculate either the fracture permeability, fracture width, or fracture conductivity. This allows ﬂexibility to input two of the variables and have the third calculated, where the calculated value will then be dimmed. The calculate option may not be available depending on the fracture options selected (i.e., if input conductivity or calculate fracture permeability is selected). In this case, fracture width and average fracture conductivity were the input, and fracture permeability was subsequently calculated. Effective dimensionless conductivity in the pay zone is then calculated from:

0 kf wf hp CfD = · (30) kL hp

The parameters from MFrac include:

• Total pay zone height [m],

• Effective propped pay zone height [m],

• Propped fracture length [m],

• Fracture width [in], and

• Average fracture conductivity [mD-ft].

In addition, the inverse fracture diffusivity is a parameter to consider as the conductivity of a fracture approaches the conductivity of the reservoir, which may significantly influence the early time production behavior of a well [72]. The definition of the inverse fracture diffusivity is: kφct cf = (31) kf φf ctf where

k = Equivalent reservoir permeability

kf = Equivalent fracture permeability

ct = Total reservoir compressbility

ctf = Total fracture compressibility φ = Equivalent reservoir porosity

φf = Equivalent fracture porosity Cari Covell 73

The calculation results in a small ratio of 10/1100 for permeability and porosity, while compressibility remains fairly constant. Therefore, the inverse fracture diffusivity is of the order 8e-05. Lastly, the fracture skin factor is assumed to be negligible (zero).

4.4.5 Well data

Production forecasting typically requires a minimum of data describing certain features of the well in order to perform a simulation. For MProd, these features include the wellbore radius and the permeability damage due to skin. The well radius deﬁnes the contact area between the well and the reservoir. The skin damage (well skin) characterizes the additional pressure drop associated with the near wellbore effects. This parameter may have a signiﬁcant effect on calculating the increase in productivity index. For the purposes of this thesis, the skin factor before stimulation was assumed to be 0.67 and after stimulation to be small and negative at approximately -0.06 as indicated by Shengtao’s reservoir analysis [61].

5.16146C CD = 2 (32) 2πcth(rw ) where

CD = dimensionless wellbore storage C = wellbore storage coefﬁcient [61]

ct = total reservoir compressibility [61] h = pay zone height [MFrac]

rw = wellbore radius [61] φ = equivalent reservoir porosity [61]

Table 14 summarizes the input parameters of well data for MProd.

Table 14: Well data for Hoffell well HF-1.

Property Value Unit Wellbore radius 0.14 [m] Wellbore skin factor (base - prefrac) 0.67 Wellbore skin factor (stimulated) -0.06 Wellbore storage factor 1.302 74 75

Chapter 5

Results

This chapter will discuss the results from MFrac and MProd, followed by a Lumpﬁt parameter model showing production potential after stimulation.

5.1 MFrac

The packer was assumed to be placed between 1070-1110 m depth to be conservative with a placement interval, with stimulation below the packer from 1110-1608 m depth in the intrusive volcanic region (total interval length 498 m), and stimulation above the packer from 400-1070 m depth in the dolomite region (total interval length 670 m). For full reports see AppendixE. Results from proppant design tabulations indicated no need to continue modeling stimulation below the packer, however all results continue to be analyzed for further discussion in Chapter6.

5.1.1 Fracture propagation solution

The fracture propagation solution displays the fracture (frac) length, height, and width characteristics. Fracture length is measured as a half-length, i.e. for one wing of the fracture. The fracture length created after stimulation below the packer (122 m) is significantly shorter than the fracture length created after stimulation above the packer (925 m), which is due to permeability of the rock type defined for each stimulation zone. Perme- ability for the intrusive volcanic zone below the packer is defined as 10−4 darcy, which is significantly less than the permeability for the dolomite zone above the packer defined as 76 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

10−1 darcy [81]. Each permeability value is based on geological surveys in other areas of Iceland similar to the Hoffell geothermal ﬁeld [81].

Frac height is calculated above and below the target fracture zone (i.e. where the packer is isolated between 1070-1110 m depth). When comparing upper frac height and lower frac height, the one that is greater is the one where injection is dominant, as seen in Figures 21 and 22. Lower frac height is greater than upper frac height when injection is below the packer, and upper frac height is greater than lower frac height when injection is above the packer.

Figure 21: Upper and lower fracture zone height when stimulated below the packer.

Figure 22: Upper and lower fracture zone height when stimulated above the packer.

Frac width is illustrated in Figures 23 and 24 as a percentage of frac length, where frac width decreases as frac length increases. The proportion that governs each value related to fracture length, height, and width is unknown; which will be addressed in Chapter6: Discussion. Cari Covell 77

Figure 23: Frac width as a function of frac length for stimulation below the packer.

Figure 24: Frac width as a function of frac length for stimulation above the packer.

Net frac pressure is total pressure measured over the 12 hour stimulation, subtracted by pressure loss in the wellbore due to friction. Effects of net frac pressure over time are illustrated in Figure 25. Net frac pressure remains fairly constant at later stages of stimulation after about 30 minutes time, but is very low at about 0.5 bar. Net frac pressure increases with time when stimulation is done above the packer, which seems abnormal upon ﬁrst glance. However this observation is consistent with pressure measured in the initial production test of Hoffell HF-1, as pressure is assumed to stabilize when measured over a longer period of time. The net frac pressure tabulated in MFrac will later be used as input for MProd, as these calculations represent pressure measured from a production test conducted after stimulation. 78 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Figure 25: Net pressure measured after stimulation below and above the packer.

The fracture propagation solution is summarized in Table 15 for stimulation below and above the packer.

Table 15: Fracture propagation solution. Calculated values are at the end of treatment.

Parameter Value Unit Below packer Above packer Length (one wing) 122.04 924.92 [m] Upper frac height 42.868 140.13 [m] Lower frac height 637.1 64.037 [m] Total frac height 679.97 204.17 [m] Avg. hydraulic frac width 0.75954 0.12945 [in] Net frac pressure 0.52949 10.604 [bar]

5.1.2 Proppant design summary

The proppant design summary is shown in Table 16. Proppant concentration is present at the end of job (EOJ) within the fracture after stimulation below the packer, suggesting that the proppant aided in creating the fracture. A different result is observed regarding proppant concentration in the pay zone as no proppant concentration is present, therefore no fracture was propped after stimulation. The proppant failed to create any fracture propagation that would allow for an increase in ﬂuid production, and the stimulation is therefore deemed unsuccessful. This is possibly due to the low concentration of 0.571429 Cari Covell 79 lbm/gal proppant in the slurry solution of 2,606 m3 (693,480 gal), compared to large amounts of proppant used in oil and gas applications.

Proppant concentration in the fracture and in the pay zone after stimulation above the packer are present in very small amounts, but parallel the fracture length created and propped. The corresponding propped height is calculated to be about 17% of the propped fracture length. The well and pay zone are also propped due to stimulation above the packer, and width of propagation is accounted for in MProd production analyses described in the next section of this chapter. Fracture conductivity in the pay zone is observed to be very low at about 12 mD-ft, foreshadowing indication that stimulation effects on production output are minor. Dimensionless frac conductivity is displayed via internal calculations in MFrac. The average fracture permeability increased from 10−1 darcy to 86.526 darcy, which indicates that stimulation was successful in terms of improving permeability of the formation.

Table 16: Proppant design summary for stimulation below and above the packer.

Parameter Value Unit Below packer Above packer Frac length - created 122.04 924.92 [m] Frac length - propped 0 92.946 [m] Propped height (pay zone) avg. 0 16.445 [m] Propped width (well) - avg. 0 0.13675 [in] Propped width (pay zone) - avg. 0 0.0016789 [in] Conc./area (frac) - avg. at EOJ 0.27101 0.040309 [lbm/ft2] Conc./area (pay zone) - avg. at closure 0 0.0014565 [lbm/ft2] Frac conductivity (pay zone) - avg. at closure 0 12.106 [mD-ft] Dimensionless frac conductivity (pay zone) 0 3.9699e-05 Avg. fracture permeability 0 86.526 [darcy]

5.1.3 Total ﬂuid loss and leakoff rate output

A summary of fluid loss volume and fluid efficiency in the fracture, for stimulation below and above the packer, is shown in Table 17. The total fluid loss corresponds to frac fluid efficiency, where over 18 times more total fluid loss is observed from stimulation above the packer than from stimulation below the packer. To better illustrate the cause of this trend, Figure 26 shows frac fluid efficiency as a function of time. Frac fluid efficiency due to stimulation below the packer stays relatively constant at 96%, where slightly more 80 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

fluid loss occurs at the initial stages of stimulation at about 86% efficiency. Frac fluid efficiency due to stimulation above the packer sees a steeper decline at the initial stages of stimulation from 82% to 69% efficiency, followed by more gradual fluid losses over the duration of the stimulation and ending at about 40% efficiency. Changes in frac fluid efficiency from the initial stages to the later stages of stimulation is dependent on the relationship between slurry volume injected and liquid volume injected, as well as the rate of fluid leakoff over the 12 hour stimulation.

Liquid volume injected for each case is rather similar as this represents the real volume injected into the wellbore during stimulation. Volume loss occurs during the first stage of stimulation (about 30 minutes long), and acts as a cleaning (or flushing) of the wellbore period before actual stimulation occurs. Fluid loss at the initial stages of stimulation below the packer are about twice as great as that above the packer, possibly due to more obstructions in the lower part of the well. Nevertheless, fluid loss at the initial stages of stimulation is most likely due to cleaning of drill cuttings blocking the well from further stimulation.

When stimulation occurs below the packer, frac fluid efficiency increases from initial stages to later stages by about 10%. The trend simply shows that less fluid loss occurs during stimulation than during well cleaning, although this seems to be a rare occurrence and is unique behavior compared to other geothermal stimulation operations; especially since fluid loss over the later stages of stimulation is only about 4% of fluid volume injected into the wellbore. This result is unrealistic because no propped fracture was created, and therefore the amount of fluid loss should be much higher. When stimulation occurs above the packer, frac fluid efficiency decreases from initial stages to later stages by about 30%. The trend is typical for a geothermal well; and the amount of fluid loss seems logical as Hoffell well HF-1 has a history of large fluid loss, as measured from leakoff tests performed after drilling. These results are consistent and in agreement with the leakoff rate illustrated in Figure 27.

Table 17: Summary of ﬂuid loss below and above the packer. Calculated values are at the end of treatment.

Parameter Value Unit Below packer Above packer Slurry volume injected 2606 2606 [m3] Liquid volume injected 2540.7 2582.2 [m3] Fluid loss volume 86.542 1551.6 [m3] Frac ﬂuid efﬁciency 0.96679 0.4046 Cari Covell 81

Figure 26: Frac ﬂuid efﬁciency after stimulation below and above the packer.

Figure 27: Leakoff rate after stimulation below and above the packer.

5.1.4 Heat transfer solution

Heat transfer effects were modeled in MFrac in order to analyze injection fluid temperature over the 12 hour stimulation. Figure 28 shows measured temperature of the frac fluid as a function of fracture length created after stimulation. In both instances, the frac fluid heats up to reservoir temperature within the first few meters over the length of the fracture. 82 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Figure 28: Temperature as a function of fracture length after stimulation below and above the packer.

5.2 MProd

Due to the case of an unsuccessful fracture propagation when stimulating below the packer, there is no need for further analysis in Mprod and Lumpﬁt. Numerical tabulations for stimulation above the packer are not validated, in terms of magnitude, because there is no prior experience in low temperature geothermal reservoir stimulation modeling to compare results. However graphical trends are justiﬁed for moving forward with analysis in the case of stimulation above the packer, and will therefore be the only case discussed for the remainder of this chapter.

In MProd single case fracture characteristics must be entered based on results from MFrac simulation, and as such are listed in Table 18 for stimulation above the packer. Perme- ability was calculated by MProd based on inputs of fracture width and average fracture conductivity. Dimensionless conductivity and inverse fracture diffusivity are internally calculated as described earlier in section 4.4.4 of the Methods chapter. The fracture skin factor is assumed to be negligible (zero). Cari Covell 83

Table 18: MProd single case fracture characteristics dialog box.

Parameter Value Unit Total pay zone height 670 [m] Effective propped pay zone height 16.445 [m] Propped fracture length 92.946 [m] Fracture permeability, kf 1122.22 [mD] Fracture width, wf 0.12945 [in] Average fracture conductivity, kfwf 12.106 [mD-ft] Dimensionless conductivity Cfd 0.00017137 Fracture skin factor 0 Inverse fracture diffusivity 8e-05

The net pressure data (in bar) obtained from MFrac is then used as a boundary condition for MProd simulation, while the initial production data of Hoffell well HF-1 is overlayed for comparison. The simulated flow rate calculated before and after stimulation over the stimulation time of 12 hours is illustrated in Figure 29. Note that MProd has a maximum capacity of 500 iterations for the overlay production data, therefore the initial set of production data measured twice a day was cut to only include measurements taken once a day; where the input was flow rate in L/s over time in days. The production test is a long- term test conducted over one year, while the stimulation is a short-term test conducted over 12 hours, therefore the flow rate calculated in MProd before stimulation has been interpolated to accommodate the short-term test. Due to this interpolation the flow rate is slightly skewed, but makes little difference regarding outcomes of the final analyses. Ad- ditionally Figure 29 shows the flow rate before stimulation to be fairly constant at 20 L/s towards the end of the 12 hour cycle, which is consistent with measurements from the first step of the initial production test. Flow rate after stimulation is then improved by some productivity index. Therefore an assumption is made that upon a long-term production test after stimulation, the flow rate will correspond in similar steps. 84 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Figure 29: Flow rate of Hoffell well HF-1 before and after stimulation.

A production ratio in relation to flow rate and to volume is then calculated for each iteration, where the results are summarized in Table 19 for approximately every 2 hours iteration (full report in AppendixF). PI based on volume is slightly higher than PI based on flow rate because the average volume is used as the initial condition. For simplification of further analyses the average productivity index of 1.096 (from flow rate) was carried forward.

Table 19: MProd production solution for Hoffell well HF-1.

Time Flow rate pre- Flow rate Cumulative Avg. reservoir Flowing Prod. ratio Prod. ratio

stimulation post-stimulation prod. pressure pressure Q/Qbase V/Vbase [d] [L/min] [L/min] [m3] [bar] [bar] [J/Jo(t)] [J/Jo|avg] 0.0023385 1797.2 2033.3 8.5725 38.5 7.4687 1.1314 1.1501 0.041997 1423.2 1572 105.81 38.5 8.2045 1.1045 1.1148 0.083469 1325.3 1457.1 195.8 38.5 8.9667 1.0995 1.1087 0.17211 1240.8 1358.6 374.69 38.5 9.6243 1.0949 1.1031 0.25468 1196.6 1307.3 532.82 38.5 10.001 1.0926 1.1003 0.33728 1166.6 1272.7 686.06 38.5 10.26 1.0909 1.0983 0.41985 1144.4 1247.1 835.75 38.5 10.452 1.0897 1.0969 0.5027 1126.7 1226.8 983.22 38.5 10.604 1.0888 1.0957

5.3 Lumpﬁt parameter model

Goals for lumpfit parameter modeling are to evaluate changes in injectivity index, storativity, transmissivity, and productivity index after stimulation of Hoffell well HF-1. After the 12 hour stimulation operation was modeled in MProd, the production flow rate im- Cari Covell 85 provement ratio of 1.096 was used to scale the initial production data of flow rate versus water level. Previous analyses indicate a two-tank open model to show good fit between measured and calculated production data before stimulation, however the same may not be true for production data after stimulation. To test the fit of post-stimulation data, Figure 30 compares lumpfit parameter models for a two-tank closed and a two-tank open reservoir over a hypothetical time frame of one year (365 days) in order to mimic a long-term production test. As anticipated the two-tank closed model fit is poor while the two-tank open model fit is good, therefore the two-tank open model is used in analyses moving forward. Production flow rate is shown in Figure 31, where increases are observed from 20 l/s to 22 l/s in the first step to day 86, and from 15 l/s to 16 l/s in the second step to day 189. The remainder of the production flow rate data stays relatively constant with only minor increases after stimulation. An average reservoir temperature of 69°C was assumed to remain constant before and after stimulation.

Figure 30: Measured and calculated water level of a long-term production test after stimulation of Hoffell well HF-1.

Figure 31: Long-term production test before and after stimulation for a one year period of well HF-1. 86 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Storativity, transmissivity, injectivity index, and productivity index are expressed in Table 20 to compare improvement ratios before and after stimulation. Tabulations were made based on the equations deﬁned in Chapter 2: Background and the assumptions made in Chapter 3: Methods. Improvement ratios are based on year-long production data before stimulation and production data after stimulation above the packer.

Table 20: Improvement ratios for well test data after stimulation above the packer.

Parameter Value Unit Improvement ratio Before stim. ﬁve Before stim. year- After stim. months prod. test long prod. test data prod. test data data [61] Storativity 5.65E-08 5.10204E-05 5.10204E-05 [m3/Pa·m2] 1 (S) Transmissivity 1.2E-08 4.59E-07 4.78E-07 [m3/Pa·s] 1.041 (T) Injectivity 3.479 1.77 1.942 [(L/s)/bar] 1.097 index (II) Productivity 0.4084 0.0378 0.0413 [(L/s)/bar] 1.093 index (PI)

Once again, future predictions of production rate could be calculated to estimate the response of the water level (reservoir pressure) to exploitation. The same scenario of using ten year predictions were calculated to help gain an understanding of the general water level changes for different production ﬂow rates [61]. Production rates were set to the same four limits based on Monte Carlo simulation of the Hoffell reservoir; 28.6 l/s, 21.4 l/s, 14.3 l/s, and 7.15 l/s [61]. Upon further analysis, an additional boundary condition of 35.15 l/s was used to better measure well output, determined by adding one more in- crement of 7.15 l/s to the previous upper boundary condition of 28.6 l/s; or essentially modeling a sustained ﬂow of 35.15 l/s to have about 7.5 MWth reservoir capacity for a lifetime of 25 years. As seen in Figure 32, the year-long analysis after stimulation shows the open system as reaching equilibrium in a similar fashion to the model made before stimulation seen in Chapter 4: Methods. Cari Covell 87

Figure 32: Predicted water levels in well HF-1 for the next 10 years for different production rates using the simulated year-long period of production test data after stimulation. The optimistic two-tank approach is shown.

Changes in water level are observed based on production ﬂow rate from initial analyses made using ﬁve-month production data before stimulation, as well as analyses made using year-long production data before stimulation and after stimulation. Results are tabulated for comparison in Table 21. With the pump at a depth of 175 m, a sustained production increase is seen from 28.6 l/s to 32.4 l/s.

Table 21: Predicted water levels after stimulation in well HF-1 after 10 years production based on year-long production data

Production ﬂow rate Optimistic model Optimistic model [L/s] (before stimulation) [m] (after stimulation) [m] 7.15 -44 -38 14.3 -88 -77 21.45 -132 -115 28.6 -176 -153 35.75 N/A -193 88 89

Chapter 6

Summary

This chapter will provide a basis for discussion regarding hydraulic stimulation in low temperature geothermal areas for direct use.

6.1 Discussion

District heating and direct use of hot water through renewable energy resources are globally in demand. Optimal methods of extraction are always under development as resource security is essential over the lifetime of a reservoir. In the past, stimulation in low temperature geothermal areas was done as part of drilling completion, especially for wells showing little to no productivity. Within low temperature geothermal areas of Iceland, great success has been seen in the Reykir hydrothermal area and in Seltjarnarnes well SN-12 on the order of 30-60 fold increases in production flow; however little success was seen in other areas such as Hlidardalur on the order of only 2-3 l/s increase in production flow. Ultimately, stimulation after drilling completion stopped due to small benefits seen given the amount of cost, and production was subsequently not adequate for a given demand. Therefore stimulation has since been translated to other high temperature and enhanced geothermal fields where more production potential has been proven based on preliminary studies.

The theory of hydraulic stimulation is associated with several methods in regards to using different types of proppants and fluids. Most geothermal stimulations have been performed using the open hole packer method, while few cases have been performed using zonal isolation. Motivation for the use of stimulation software is based on the need to compare several types of hydraulic stimulation methods along with testing different com- 90 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use binations of proppant concentration and volume of fluid injected. Currently, FRACPRO software is associated with modeling fracture geometry after stimulation, but not production. Other fracture modeling software is available, but in geothermal applications has mostly been used to analyze seismic effects from stimulation to moderate environmental impact. Based on these findings, a solution software was sought through the oil and gas industry. Research in Canada on the tight sandstone EGS environment for extracting oil has led to the use of MFrac suite software. MFrac was used to model a hypothetical stimulation, and the sub-program MProd was used to model production after stimulation. The MFrac suite is designed to model hydraulic stimulation based on geological characteristics and proppant fluid (frac fluid), while also including solutions for fluid losses due to leakoff and friction. This is in order to create a production flow rate simulation after a hydraulic stimulation operation.

MFrac was then implemented for the case study of Hoffell well HF-1, via open hole stimulation using two scenarios below and above a packer. The auto-design treatment schedule was used as not enough information is available to conclude best frac fluid design within stimulation practices in low temperature geothermal areas in Iceland. In MFrac, additional assumptions were necessary as the software is designed for oil and gas applications; in order to model well HF-1 as accurately as possible. For example, MFrac is designed for stimulation via the use of perforations within a liner similar to a zonal isolation technique. Geothermal stimulation in Iceland is typically performed using an open hole packer, therefore modifications in the wellbore hydraulics dialog box were necessary to take into consideration. Another factor for consideration was the placement of the packer, and subsequently the intervals of stimulation below and above the packer. No specific guideline within the MFrac software was provided to accommodate a packer, which is odd as packers are very common in the oil and gas industry.

Upon MFrac simulation, figures showed very long fractures with very short fracture heights. These results were anticipated because of the water frac fluid used, however the proportion seems unrealistic compared to other fracture simulations done using FRACPRO in Gross Schonbeck, Germany. One indication regarding this outcome includes permeability influence from the rock type assumed. Basaltic areas in Iceland are more permeable than the tight sands of EGS sites, such as Soultz and Gross Schonbeck in the Upper Rhine Grabbon of Europe. However, typical fracture behavior in Iceland after stimulation is unknown, especially in low temperature geothermal areas. When propped fracture length and height were analyzed, MFrac indicated an inconclusive result for stimulation below the packer. The reason for no fracture propagation below the packer could be due to the low concentration of proppant used. The oil and gas industry typically uses large amounts of proppant in stimulations, therefore MFrac software is most likely not sensitive Cari Covell 91 enough to a very water based frac fluid. Stimulation intervals are much larger in geothermal applications than in oil and gas applications, so this could also be an effect from the sensitivity of MFrac. After some test runs, results showed that increasing concentration with respect to defining a shorter interval indicated more fracture propagation, but not necessarily through the entire length of the created fracture.

For MProd, most of the input parameters are accurately assumed for the case study of Hoffell well HF-1. Therefore MProd is considered a reliable simulation software based on initial reservoir conditions to show what the improvement ratio would be after stimulation. However, MProd created interpolations of the long-term production test data to accommodate for the short-term stimulation, which produces some minor inaccuracies. The desired result of finding an improvement ratio after stimulation was deemed fairly accurate for the case of Hoffell well HF-1. The solution methodology used in MProd to create the final flow rate tabulations needs further analysis, discussed further in the conclusions section 6.2.

For lumpfit parameter modeling (LPM), an increase in flow was observed but only for the optimistic open tank system, therefore the overall improvement factor for well HF-1 is hard to determine accurately. A better fit for the closed tank model would indicate the effects of stimulation. Perhaps from the assumptions made earlier, a 15-20 l/s rate would increase to somewhere around 20-25 l/s, but this increase would most likely be small of no more than 5 l/s. Therefore well HF-1 is not considered to be a good candidate for stimulation, as improvement ratios on the order of 2-3 fold increase have been seen throughout some wells in the Reykir hydrothermal field in Iceland. However, the methodology can be applied to other geothermal areas around the world, since this thesis is the first to initiate stimulation predictions for low temperature geothermal applications.

6.2 Conclusions

Stimulation above the packer was successfully modeled in MFrac and MProd, while stimulation below the packer had inconclusive results for the case study of Hoffell well HF-1. However, this does not mean that stimulation above the packer is a better method for use across other geothermal areas as there are several uncertainties about the case study. The productivity improvement was minor with only a 1.096 improvement ratio, therefore the well is not a good candidate for realistic stimulation operations. In addition, very optimistic production flow rates were measured in the open tank geothermal system model, which further indicates that a different candidate well would have been more appropri- 92 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use ate for analyses moving forward. Although the production flow is very optimistic, there is potential for even more flow improvement if fracture propagation were modeled more accurately in MFrac. More research is needed to understand the schematic of fracture propagation in low temperature geothermal areas of Iceland. Nevertheless, the methodology used in this thesis will be applicable and beneficial to other case studies with more data and field history.

Sensitivities were observed within MFrac, MProd, and Lumpfit beta software. These include MFrac’s proppant concentration and stimulation interval, MProd’s solution methodology for matching short-term and long-term production test data, and Lumpfit beta’s correlation with closed tank geothermal systems using long-term production test data. In- ternal programming work within MFrac, MProd, and Lumpfit beta software is necessary to account for varying geothermal circumstances. Specific to Lumpfit beta, the software is meant to be used when there is limited data available, however it was used in this thesis to compare original models done by Shengtao for the case study of well HF-1 in the Hoffell low temperature geothermal field [61]. Perhaps a different program capable of predicting reservoir properties using long-term production data would have been more applicable to use moving forward with this case study. Ultimately, this thesis initiates first attempts in modeling stimulation for direct use; showing there is proof of concept.

6.3 Future work

Several applications for future work specific to MFrac and MProd are available in addition to working on the software for geothermal applications. For MFrac, a treatment schedule can be created rather than having one auto-designed in order to look at a more realistic stimulation. Also the stimulation can be modeled over a longer period of time of a day or two, as some stimulation operations last this long or longer in order to better improve the production flow of the well. In addition experiments can be done using different proppant types, although sand is common for geothermal applications, but changing the grain size might have an effect on fracture propagation. Use for polymers in fluid solutions may also be an area of interest for further studies, in particular to polymers that are effective in scaling mitigation. Within MFrac, more complex scenarios of stimulation through multiple rock layers can be done, as well as stimulation of more than one fracture. These scenarios mimic a more realistic zonal isolation for purposes of connecting fractures to the main fault line of a geothermal area. Cari Covell 93

For MProd, history matching can be done similarly to lumpﬁt parameter modeling, where measured and calculated data simulate to look at optimization of production ﬂow. How- ever in order for this to be useful, prior stimulations must have been performed in order to establish a proper boundary condition. Lastly, fracture optimization is a separate form of MProd that can be looked into further, but this is more for the distant future.

Additional software within the MFrac suite are very useful for modeling effects of stimulation. MView is a data handling system and display module for the real-time and replay analysis of hydraulic fracture treatments, and is used widely for commercial applications in the oil and gas industry [72]. When stimulation is underway, field operators can see the fracture creation and propagation trends as they are happening, and can therefore adjust the pumping schedule and proppant concentration accordingly. In addition, MNpv is an economic analysis software, using treatment advantages or disadvantages that can be determined by evaluating predicted cash flow and future return on investment [72]. Although there are many factors which influence the results of a design, the principal objective of any stimulation operation is to maximize well profitability. Typically, this involves careful economic analysis of the costs and potential benefits of individual well operations. For many, forecasting net present value (NPV) has become an integral part of the preferred methodology to optimize hydraulic stimulation treatments [72]. The integrated capability of MFrac, MProd, and MNpv provides the ability to optimize the potential propped fracture length of a design based on the pumping schedule, fracture geometry, treatment cost, and production revenue. The NPV of various fracture lengths can then be calculated and displayed as a function of producing time. Hydraulic stimulation optimization is therefore a basic requirement to maximize economic returns on investment. the economics of price of hot water versus investment can then be compared to other sources of heating, such as coal, oil, gas, and electricity.

6.4 Recommendations

As much work can be done, the author has recommendations for moving forward with future studies. Starting with a focus in Iceland, first would be to conduct fracture modeling of a low temperature geothermal area in order to analyze typical characteristics, if such exist. Next would be to identify better candidates for stimulation, starting with areas where reinjection is a more common practice; as some wells within these areas have already been classified as having poor production. The way to identify a better candidate for stimulation is to create a lumpfit model using production test data, and properly comparing assumptions of a closed geothermal system and an open geothermal system; 94 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use in order to provide a better foundation for further stimulation analysis. MFrac and MProd are the best software to use at this time when looking at overall production after stimulation, but for fracture modeling it is recommended to use FRACPRO as the software is more accommodating to geothermal conditions. Although Hoffell well HF-1 did not turn out to be a good candidate for a case study, interests lye in how zonal isolation techniques would work in this area, as televiewer logs indicated need to connect to the main fault line. However this requires more knowledge on stimulation of multiple fractures through multiple layers of rock.

In a broader schematic, the methodology conducted throughout this thesis is recommended for applications in high temperature and EGS areas. While fracture modeling has been the focus of much research, it is equally as important to evaluate production flow via a hydraulic stimulation model. Preliminary work regarding economic analyses for stimulation in high temperature and EGS areas is also recommended; which includes gathering information on best types of proppant to use for particular geothermal areas, and current costs of stimulation equipment and frac fluids. The author plans to continue research on hydraulic stimulation in low temperature geothermal areas specific to Iceland, while also considering other low temperature geothermal sites throughout Europe and the United States, in order to contribute towards exploring renewable solutions for meeting direct use demand in local communities around the world. 95

References

[1] M. H. Dickson and M. Fanelli, Geothermal energy: utilization and technology. Routledge, 2013.

[2] J. W. Tester, E. M. Drake, M. J. Driscoll, M. W. Golay, and W. A. Peters, Sustain- able energy: choosing among options. MIT press, 2005.

[3] J.-C. Roegiers, R. Jeffrey, G. Axelsson, K. Evans, and T. Powell, “White Pa- per: Stimulation Procedures,” International Partnership for Geothermal Technol- ogy, Tech. Rep., August 2012.

[4] G. Axelsson and S. Thórhallsson, “Review of well stimulation operations in Ice- land,” Transactions, v33, Geothermal Resources Council, 2009.

[5] I. Stober, Tiefe Geothermie: Nutzungsmöglichkeiten in Deutschland. Bundesmin für Umwelt, Naturschutz und Reaktorsicherheit (BMU), 2011.

[6] M. Grant, “Thermal stimulation: the easiest and least recognised mechanism of permeability change,” in United States/New Zealand Geothermal Workshop, 16-20 April 2012.

[7] M. J. Economides, K. G. Nolte, U. Ahmed, and D. Schlumberger, Reservoir stimulation. Wiley Chichester, 2000, vol. 18.

[8] J. Tittman and J. Wahl, “The physical foundations of formation density logging (gamma-gamma),” Geophysics, vol. 30, no. 2, pp. 284–294, 1965.

[9] I. D. Palmer, R. W. Veatch Jr et al., “Abnormally high fracturing pressures in step- rate tests,” Society of Petroleum Engineers Production Engineering, vol. 5, no. 03, pp. 315–323, 1990.

[10] L. Harrington, N. Whitsitt, and R. Hannah, “Prediction of the Location and Move- ment of Fluid Interfaces in a Fracture,” Southwestern Petroleum Short Course, Texas Tech University, Lubbock, Texas, USA, Tech. Rep., 1973. 96 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

[11] K. G. Nolte and M. B. Smith, “Interpretation of fracturing pressures,” Journal of Petroleum Technology, vol. 33, no. 09, pp. 1–767, 1981.

[12] A. Aqui and S. Zarrouk, “Permeability Enhancement of Conventional Geother- mal Wells,” in Proceedings New Zealand Geothermal Workshop, Auckland, New Zealand, 21-23 November 2011.

[13] D. Vollmar, V. Wittig, and R. Bracke, “Geothermal Drilling Best Practices: The Geothermal translation of conventional drilling recommendations-main potential challenges,” International Geothermal Association: Home, 2013.

[14] A. Reinicke, E. Rybacki, S. Stanchits, E. Huenges, and G. Dresen, “Hydraulic fracturing stimulation techniques and formation damage mechanisms—Implications from laboratory testing of tight sandstone–proppant systems,” Chemie der Erde- Geochemistry, vol. 70, pp. 107–117, 2010.

[15] G. Axelsson, “Geothermal Well Testing,” United Nations University, 2013.

[16] M. Flores, D. Davies, G. Couples, and B. Palsson, “Stimulation of geothermal wells, can we afford it?” in Proceedings World Geothermal Congress, Antalya, Turkey, 24-29 April, 2005.

[17] M. A. Grant and P. F. Bixley, Geothermal Reservoir Engineering. Academic Press, 2011.

[18] H. Hédinsdóttir, “Mechanisms of injectivity enhancement in the thermal stimulation of geothermal wells,” Master’s thesis, Swiss Federal Institute of Technology Zurich, Department of Earth Sciences, August 3 2014.

[19] G. Axelsson, S. Thórhallsson, and G. Björnsson, “Stimulation of geothermal wells in basaltic rock in Iceland,” in Enhanced geothermal innovative network for Europe Workshop, vol. 3, 2006.

[20] L. Kalfayan, Production enhancement with acid stimulation. Pennwell Books, 2008.

[21] J. J. Mortensen, “Hot dry rock: a new geothermal energy source,” Energy, vol. 3, no. 5, pp. 639–644, October 1978.

[22] J. Tomasson and T. Thorsteinsson, “Use of injection packer for hydrothermal drillhole stimulation in Iceland,” in Second UN Symposium on the Development and Use of Geothermal Resources, San Francisco, 1975. Cari Covell 97

[23] ENGINE Coordination Action, “Best Practice Handbook for the development of Unconventionnal Geothermal Ressources with a focus on Enhanced Geothermal System,” Collection Actes/Proceedings, 2008.

[24] J. Walters, S. Thorhallsson, and E. Wood, “White Paper: Zonal Isolation for Geothermal Wells,” International Partnership for Geothermal Technology, Tech. Rep., August 2012.

[25] G. Andersen, S. Pye, and A. Probst, “Gas lift process for restoring ﬂow in depleted geothermal reservoirs,” Patent US4 787 450 A, 29 November 1988. [Online]. Available: https://www.google.com/patents/US4787450

[26] F. G. Driscoll, “Groundwater and wells,” St. Paul, Minnesota: Johnson Filtration Systems Inc., 1986, 2nd ed., vol. 1, 1986.

[27] Australian Drilling Industry Training Committee, Drilling: the manual of methods, applications, and management. Lewis Pub., 1997.

[28] A. Shadravan, M. Ghasemi, and M. Alﬁ, “Zonal Isolation in Geothermal Wells,” in Proceedings, Fortieth Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, California, 26-28 January 2015.

[29] D. L. Marcus and C. Bonds, “Results of the reactant sand-fracking pilot test and implications for the in situ remediation of chlorinated VOCs and metals in deep and fractured bedrock aquifers,” Journal of hazardous materials, vol. 68, no. 1, pp. 125–153, 1999.

[30] E. L. Majer, R. Baria, M. Stark, S. Oates, J. Bommer, B. Smith, and H. Asanuma, “Induced seismicity associated with enhanced geothermal systems,” Geothermics, vol. 36, no. 3, pp. 185–222, 2007.

[31] E. Gaucher, M. Schoenball, O. Heidbach, A. Zang, P. Fokker, J.-D. van Wees, and T. Kohl, “Induced Seismicity in Geothermal Reservoirs: Physical Processes and Key Parameters,” in Proceedings World Geothermal Congress, Melbourne, Australia, 19-24 April 2015.

[32] M. O. Häring, U. Schanz, F. Ladner, and B. C. Dyer, “Characterisation of the Basel 1 Enhanced Geothermal System,” Geothermics, vol. 37, no. 5, pp. 469–495, 2008.

[33] F. Cornet, Y. Jianmin, and L. Martel, “Stress heterogeneities and ﬂow paths in a granite rock mass,” in Pre-Workshop Volume for the Workshop on Induced Seis- micity, 33rd U.S. Symposium on Rock Mechanics, 1992, p. 184. 98 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

[34] F. Cornet and O. Scotti, “Analysis of induced seismicity for fault zone identiﬁca- tion,” in International journal of rock mechanics and mining sciences & geomechanics abstracts, vol. 30, no. 7. Elsevier, 1993, pp. 789–795.

[35] F. Cornet and P. Julien, “Stress determination from hydraulic test data and focal mechanisms of induced seismicity,” in International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, vol. 26, no. 3. Elsevier, 1993, pp. 235–248.

[36] J. Brune and W. Thatcher, “International Handbook of Earthquake and Engineering Seismology, vol. 81A,” International Association of Seismology and Physics of Earth’s Interior, Committee on Education, pp. 569–588, 2002.

[37] R. C. M. Malate, “Well intervention techniques,” 2003.

[38] T. Chu, R. Jacobson, and N. Warpiniski, “Geothermal well stimulation using high energy gas fracturing,” in Proceedings of Twelfth workshop on Geothermal Reser- voir Engineering, Stanford University, Stanford, California, 1987, pp. 20–22.

[39] M. Batyrbaev, V. Bulavin, V. Seljakov, and A. Savchenko, “Application of Tech- nology of Electroinﬂuence for Intensiﬁcation of an Oil Recovery in Russia and Abroad,” Neftyanoe Khozyaistvo - Oil Industry, vol. 2002, no. 11, pp. 92–95.

[40] H. Kristmannsdóttir, S. E. Stefánsson, and A. Björnsson, “Sustainable Geothermal Production of the Seltjarnarnes Geothermal Field, SW-Iceland,” in Proceedings World Geothermal Congress, Melbourne, Australia, 19-25 April 2015.

[41] H. Tulinius, G. Axelsson, J. Tomasson, H. Kristmannsdottir, and A. Gudmunds- son, “Stimulation of well SN-12 in the Seltjarnarnes low-temperature ﬁeld in SW- Iceland,” in Proceedings of the 21st Workshop on Geothermal Reservoir Engineer- ing, Stanford University, Stanford, California, 1996, pp. 489–496.

[42] T. Thorsteinsson, “Redevelopment of the Reykir hydrothermal system in southwestern Iceland,” in Second UN Symposium on the Development & Use of Geother- mal Resources, San Francisco, 1975.

[43] T. Thorsteinsson and J. Tomasson, “Drillhole stimulation in Iceland,” Am. Soc. Mech. Eng.,(Pap.);(United States), vol. 78, no. CONF-781112-, 1979.

[44] H. Tulinius, A. Spencer, G. Bodvarsson, H. Kristmannsdottir, T. Thorsteinsson, and A. Sveinbjornsdottir, “Reservoir studies of the Seltjarnarnes geothermal ﬁeld, Iceland,” Lawrence Berkeley Lab CA (USA); Iceland National Energy Authority, Cari Covell 99

Reykjavik Geothermal Div.; Iceland Univ., Reykjavik Science Inst., Tech. Rep., 1986.

[45] H. Kristmannsdottir, “Injection of drillholes at Laugarland, Eyjafjordur,” National Energy Authority of Iceland (In Icelandic), Tech. Rep. OS-JHD-7719, June 1977.

[46] J. Combs, S. K. Garg, and J. W. Pritchett, “Geothermal Well Stimulation Technol- ogy: A Preliminary Review,” Geothermal Resources Council Transactions, vol. 28, pp. 207–212, 2004.

[47] K. Yoshioka, J. Jermia, R. Pasikki, and A. Ashadi, “Zonal Hydraulic Stimulation in the Salak Geothermal Field,” in Proceedings World Geothermal Congress, Mel- bourne, Australia, 19-24 April, 2015.

[48] D. J. Entingh, “Geothermal well stimulation experiments in the United States,” in Proceedings World Geothermal Congress, Kyushu-Tohoku, Japan, May 28 - June 10, 2000.

[49] C. Morris and M. Bunyak, “Fracture stimulation experiments at the Baca Project Area,” in Proceedings 7th Workshop on Geothermal Reservoir Engineering, Stan- ford, California, USA, December 1981, pp. 53–60.

[50] A. Barelli, G. Cappetti, G. Manetti, and A. Peano, “Well stimulation in Latera ﬁeld,” Geothermal Resources Council Transactions, vol. 9, pp. 213–219, 1985.

[51] R. G. Pasikki, F. Libert, K. Yoshioka, and R. Leonard, “Well stimulation techniques applied at the Salak geothermal ﬁeld,” in Proceedings World Geothermal Congress, Bali, Indonesia, vol. 25, 2010, p. 29.

[52] M. Esberto and Z. Sarmiento, “Numerical modeling of the Mt. Apo geothermal reservoir,” in PROCEEDINGS, Twenty-Fourth Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, California, 25-27 January, 1999.

[53] M. Malibiran, J. Leyrita, B. Sambrano, and J. Nogara, “Well permeability enhancement through hydraulic stimulation conducted in the Mt. Apo geothermal project,” in PROCEEDINGS, Thirty-Eighth Workshop on Geothermal Reservoir Engineer- ing, Stanford University, Stanford, California, February 11-13, 2013.

[54] A. Aqui and S. Zarrouk, “Permeability Enhancement of Conventional Geothermal Wells,” in Proceedings, New Zealand Geothermal Workshop, 21-23 November, 2011. 100 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

[55] D. M. G. On and R. P. Andrino Jr., “Evaluation of Hydraulic Stimulation-Induced Permeability Enhancement,” in Proceedings World Geothermal Congress, Mel- bourne, Australia, 19-25 April, 2015.

[56] K. Breede, K. Dzebisashvili, X. Liu, and G. Falcone, “A systematic review of enhanced (or engineered) geothermal systems: past, present and future,” Geothermal Energy, vol. 1, no. 4, pp. 1–27, 2013.

[57] M. Scali, M. Cei, S. Tarquini, and P. Romagnoli, “The Larderello–Travale and Amiata Geothermal Fields: Case Histories of Engineered Geothermal Systems since Early 90’s,” in European Geothermal Congress, Pisa, Italy, 3-7 June, 2013.

[58] G. Zimmermann, I. Moeck, and G. Blöcher, “Cyclic waterfrac stimulation to develop an enhanced geothermal system (EGS) - conceptual design and experimental results,” Geothermics, vol. 39, no. 1, pp. 59–69, 2010.

[59] H. Hofmann, T. Babadagli, and G. Zimmermann, “Hot water generation for oil sands processing from enhanced geothermal systems: Process simulation for different hydraulic fracturing scenarios,” Applied Energy, vol. 113, pp. 524–547, 2014.

[60] B. Meyer, “Three-dimensional hydraulic fracturing simulation on personal com- puters: theory and comparison studies,” in Society of Petroleum Engineers Eastern Regional Meeting. Society of Petroleum Engineers, 1989.

[61] L. Shengtao, “Production capacity assessment of the Hoffell low-temperature geothermal system, SE-Iceland,” United Nations University, 2014.

[62] B. Steingrimsson and A. Gudmundsson, “Geothermal borehole investigations during and after driling,” in Proceedings at Workshop on Decision Makers on Geother- mal Projects in Central America, El Salvador, 14-18 November, 2005, pp. 1–10.

[63] M. Masum, “Geothermal gradient and geology of Hoffell low-temperature ﬁeld, SE-Iceland,” United Nations University, 2014.

[64] A. Hjartarson, F. Flóvenz, and M. Ólafsson, “Probability of geothermal resources near Hoffell and Midfell in Nesjar: research and material examined,” ÍSOR- 2012/002 (in Icelandic), Tech. Rep., 2012.

[65] Stapi Geological Services, “Geothermal exploration in East-Skaftafellssýsla in the years 1993-1994,” County committee of East-Skaftafellssýsla, 1994. Cari Covell 101

[66] S. Kristinsson, H. Helgadóttir, H. Stefánsson, H. Tryggvason, F. Pétursson, and M. Ólafsson, “Drilling well HF-1: drilling history, geology and capacity measurements,” ÍSOR-013/030 (in Icelandic), Tech. Rep., 2013.

[67] H. Sarak, M. Onur, and A. Satman, “Lumped-parameter models for low- temperature geothermal ﬁelds and their application,” Geothermics, vol. 34, no. 6, pp. 728–755, 2005.

[68] S. Scholtysik, “The Role of Uncertainty for Lumped Parameter Modeling of Low Temperature Geothermal Resources,” in Fourtieth Workshop on Geothermal Reser- voir Engineering, Stanford University, Stanford, California, USA, 2015.

[69] G. Axelsson, “Simulation of pressure response data from geothermal reservoirs by lumped parameter models,” in Fourteenth Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, California, USA, 1989.

[70] G. Axelsson, G. Björnsson, and J. E. Quijano, “Reliability of lumped parameter modeling of pressure changes in geothermal reservoirs,” in Proceedings World Geothermal Congress, Antalya, Turkey, 24-29 April, 2005.

[71] G. Axelsson and P. Arason, “LUMPFIT, automated simulation of pressure changes in hydrological reservoirs. Version 3.1, user’s guide,” 1992.

[72] B. H. Incorporated, MFrac Suite 10 Hydraulic Fracturing Software User’s Guide, 2015.

[73] B. Meyer, “Design formulae for 2-D and 3-D vertical hydraulic fractures: Model comparison and parametric studies,” in Society of Petroleum Engineers Unconven- tional Gas Technology Symposium, Louisville, KY. Society of Petroleum Engi- neers, 18-21 May, 1986.

[74] B. Meyer, G. Cooper, and S. Nelson, “Real-time 3-D hydraulic fracturing simulation: theory and ﬁeld case studies,” in Society of Petroleum Engineers Annual Technical Conference and Exhibition, New Orleans, LA. Society of Petroleum Engineers, 23-26 September 1990.

[75] B. R. Meyer and M. Hagel, “Simulated mini-frac analysis,” Journal of Canadian Petroleum Technology, vol. 28, no. 05, June 1988.

[76] M. Hagel and B. Meyer, “Utilizing mini-frac data to improve design and production,” Journal of Canadian Petroleum Technology, vol. 33, no. 03, 1994.

[77] S. Árnadóttir, F. Pétursson, and H. r. Stefánsson, “Hole Oscilloscopes Measurement hole HF-1 Hoffell in Nesjar,” ÍSOR-013/025 (in Icelandic), Tech. Rep., 2013. 102 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

[78] R. G. Keck, W. L. Nehmer, and G. S. Strumolo, “A new method for predicting friction pressures and rheology of proppant-laden fracturing ﬂuids,” Society of Petroleum Engineers Production Engineering, vol. 7, no. 01, pp. 21–28, 1992.

[79] P. S. Virk, “Drag reduction fundamentals,” AIChE Journal, vol. 21, no. 4, pp. 625– 656, 1975.

[80] H. Schlichting, “Boundary Layer Theory,” McGraw-Hill, New York, 1979.

[81] B. C. Haimson and B. Voight, Crustal stress in Iceland. Springer, 1977.

[82] R. C. Earlougher Jr, “Advances in well test analysis, Henry L,” Doherty series, Monograph, Society of Petroleum Engineers, vol. 5, 1977.

[83] S. Lee and J. Brockenbrough, “A New Analytical Solution for Finite Conductiv- ity Vertical Fractures with Real Time and Laplace Space Parameter Estimation,” Society of Petroleum Engineers, vol. 12013, October 1983.

[84] S.-T. Lee, J. R. Brockenbrough et al., “A new approximate analytic solution for ﬁnite-conductivity vertical fractures,” Society of Petroleum Engineers Formation Evaluation, vol. 1, no. 01, pp. 75–88, 1986.

[85] H. Stehfest, “Numerical inversion of Laplace transforms,” Communications of the ACM, vol. 13, pp. 47–49, 1970.

[86] H. Ramey and W. M. Cobb, “A general pressure buildup theory for a well in a closed drainage area,” JPT, December 1971.

[87] B. R. Meyer, R. H. Jacot et al., “Pseudosteady-state analysis of ﬁnite conductivity vertical fractures,” in Society of Petroleum Engineers Annual Technical Conference and Exhibition. Society of Petroleum Engineers, 2005.

[88] H. Mukherjee, M. J. Economides et al., “A parametric comparison of horizontal and vertical well performance,” Society of Petroleum Engineers Formation Evalu- ation, vol. 6, no. 02, pp. 209–216, 1991.

[89] M. Soliman, J. Hunt, and W. El Rabaa, “On fracturing horizontal wells,” Society of Petroleum Engineers, vol. 18542, pp. 1–4, 1988.

[90] A. C. Gringarten, H. J. Ramey Jr, R. Raghavan et al., “Unsteady-state pressure distributions created by a well with a single inﬁnite-conductivity vertical fracture,” Society of Petroleum Engineers Journal, vol. 14, no. 04, pp. 347–360, 1974.

[91] V. Stefánsson, Á. Gudmundsson, B. Steingrímsson, G. Halldórsson, H. Ármanns- son, H. Franzon, and T. Hauksson, “Kraﬂa - well KJ-14. Drilling, research and pro- Cari Covell 103

duction characteristics (in Icelandic),” Orkustofnun, Tech. Rep. OS82061/JHD09, 1982.

[92] G. Bjornsson, “Reservoir conditions at 3-6 km depth in the Hellisheidi geothermal ﬁeld, SW-Iceland, estimated by deep drilling, cold water injection and seismic monitoring,” in PROCEEDINGS, Twenty-Ninth Workshop on Geothermal Reser- voir Engineering, Stanford University, Stanford, California, January 26-28, 2004.

[93] S. Jonsson, A. Gautason, . Guðmundsson, K. Egilsson, and G. Gunnarsson, “Drilling of production section from 933 m to 2808 m (in Icelandic),” Tech. Rep. ISOR-2003/007, 2003.

[94] A. Seifu, “Evaluation of recent temperature and pressure data from wells in Ten- daho geothermal ﬁeld, Ethiopia and from well HG-1 at Hágöngur, Iceland,” United Nations University, 2004.

[95] B. Bendall, R. Hogarth, H. Holl, A. McMahon, A. Larking, and P. Reid, “Australian Experiences in EGS Permeability Enhancement–A Review of 3 Case Studies,” in Thirty-Ninth Workshop on Geothermal Reservoir Engineering, Stanford Univer- sity, California, 2014.

[96] J. W. Tester, B. Anderson, A. Batchelor, D. Blackwell, R. DiPippo, E. Drake, J. Garnish, B. Livesay, M. C. Moore, K. Nichols et al., “The future of geothermal energy: Impact of enhanced geothermal systems (EGS) on the United States in the 21st century,” Massachusetts Institute of Technology, 2006.

[97] R. Hogarth, H. Holl, and A. McMahon, “Flow Testing Results from Habanero EGS Project,” in Proceedings, Australian Geothermal Energy Conference, Bris- bane, Australia, 14-15 November, 2013.

[98] A. McMahon and S. Baisch, “Seismicity Associated with the Stimulation of the Enhanced Geothermal System at Habanero, Australia,” in Proceedings World Geothermal Congress, Melbourne, Australia, 19-24 April, 2015.

[99] G. Meyer, A. Larking, R. Jeffrey, and A. Bunger, “Olympic Dam EGS Project,” in Proceedings World Geothermal Congress, Bali, Indonesia, 25-29 April 2010.

[100] G. Klee, A. Bunger, G. Meyer, F. Rummel, and B. Shen, “In situ stresses in borehole blanche-1/south Australia derived from breakouts, core discing and hydraulic fracturing to 2 km depth,” Rock mechanics and rock engineering, vol. 44, no. 5, pp. 531–540, 2011. 104 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

[101] F. Rummel, “Fracture mechanics approach to hydraulic fracturing stress measurements,” in Fracture mechanics of rock. Academic Press London, 1987, pp. 217– 239.

[102] P. Reid, L. McAllister, and M. Messeiller, “Status of the Paralana 2 hydraulic stimulation program,” in Proceedings Australian Geothermal Energy Conference, 2010.

[103] P. Reid, M. Messeiller, E. Llanos, and M. Hasting, “Paralana 2–well testing and stimulation,” in Proceedings Australian Geothermal Energy Conference, 2011.

[104] M. Hasting, J. Albaric, V. Oye, P. Reid, M. Messeiller, E. Llanos, P. Malin, E. Shalev, M. Hogg, and M. Alvarez, “Real-Time Induced Seismicity Monitor- ing During Wellbore Stimulation at Paralana-2 South Australia,” in Proceedings Australian Geothermal Energy Conference, 2011.

[105] J. Willis-Richards, A. Green, and A. Jupe, “A comparison of HDR geothermal sites,” in Proceedings World Geothermal Congress, Florence, Italy, 18-31 May 1995.

[106] R. Baria, J. Baumgärtner, A. Gérard, and R. Jung, “European Hot Dry Rock geothermal research programme 1996–1997,” European Commission Final Report (DGXII) EUR15925ER, 1999.

[107] M. Schindler, P. Nami, R. Schellschmidt, D. Teza, and T. Tischner, “Summary of hydraulic stimulation operations in the 5 km deep crystalline HDR/EGS reservoir at Soultz-sous-Forêts,” in Proceedings, 33rd Workshop on Geothermal Reservoir Engineering, Stanford, California, USA, 2008.

[108] M. Schindler, J. Baumgärtner, T. Gandy, P. Hauffe, T. Hettkamp, H. Menzel, P. Pen- zkofer, D. Teza, T. Tischner, and G. Wahl, “Successful hydraulic stimulation techniques for electric power production in the Upper Rhine Graben, Central Europe,” in Proceedings World Geothermal Congress, Bali, Indonesia, 25-29 April, 2010.

[109] M. Adelinet, C. Dorbath, M. Calò, L. Dorbath, and M. Le Ravalec, “Crack Features and Shear-Wave Splitting Associated with Fracture Extension during Hydraulic Stimulation of the Geothermal Reservoir in Soultz-sous-Forêts,” Oil & Gas Science and Technology–Revue d’IFP Energies nouvelles, 2015.

[110] S. Held, A. Genter, T. Kohl, T. Kölbel, J. Sausse, and M. Schoenball, “Economic evaluation of geothermal reservoir performance through modeling the complexity of the operating EGS in Soultz-sous-Forêts,” Geothermics, vol. 51, pp. 270–280, 2014. Cari Covell 105

[111] U. Schanz, H. Stang, H. Tenzer, G. Homeier, M. Hase, S. Baisch, R. Weidler, A. Macek, and S. Uhlig, “Hot dry rock project Urach - a general overview,” in Proceedings of the European Geothermal Conference, Szeged, Hungary, 2003, pp. 25–30.

[112] S. Baisch, R. Weidler, R. Voeroes, H. Tenzer, and D. Teza, “Improving hydraulic stimulation efﬁciency by means of real-time monitoring,” in Proceedings of the Twenty-Ninth Workshop on Geothermal Reservoir Engineering, Stanford Univer- sity, Stanford, California, January 26-28, 2004.

[113] T. Tischner, S. Krug, E. Pechan, A. Hesshaus, R. Jatho, M. Bischoff, and T. Wonik, “Massive hydraulic fracturing in low permeable sedimentary rock in the GeneSys project,” in Proceedings, Thirty-Eighth Workshop on Geothermal Reservoir Engi- neering, Stanford University, Stanford, California, 11-13 February, 2013.

[114] C. Schrage, C. Bems, H. Kreuter, S. Hild, and S. Volland, “Overview of the enhanced geothermal energy project in Mauerstetten, Germany,” 2012.

[115] G. Zimmermann, T. Tischner, B. Legarth, and E. Huenges, “Pressure-dependent production efﬁciency of an enhanced geothermal system (EGS): stimulation results and implications for hydraulic fracture treatments,” in Rock Physics and Natural Hazards. Springer, 2009, pp. 1089–1106.

[116] G. Zimmermann, A. Reinicke, G. Blöcher, H. Milsch, D. Gehrke, H. Holl, I. Moeck, W. Brandt, A. Saadat, and E. Huenges, “Well path design and stimulation treatments at the geothermal research well GtGrSk4/05 in Groß Schönebeck,” in Proceedings Thirty-Second Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, California, January 30-Feburary 1, 2007.

[117] G. Zimmermann and A. Reinicke, “Hydraulic stimulation of a deep sandstone reservoir to develop an Enhanced Geothermal System: Laboratory and ﬁeld experiments,” Geothermics, vol. 39, no. 1, pp. 70–77, 2010.

[118] Y. Hori, K. Kitano, H. Kaieda, and K. Kiho, “Present status of the Ogachi HDR Project, Japan, and future plans,” Geothermics, vol. 28, no. 4, pp. 637–645, 1999.

[119] D. Swenson, R. Schroeder, N. Shinohara, and T. Okabe, “Analyses of the Hi- jiori long term circulation test,” in Proceedings of the Twenty-Fourth Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, California, 25-27 January, 1999. 106 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

[120] H. Kaieda, H. Ito, K. Kiho, K. Suzuki, H. Suenaga, and K. Shin, “Review of the Ogachi HDR project in Japan,” in Proceedings of the World Geothermal Congress, Antalya, Turkey, 24-29 April, 2005.

[121] K. Kitano, Y. Hori, and H. Kaieda, “Outline of the Ogachi HDR project and charac- ter of the reservoirs,” in Proceedings World Geothermal Congress, Kyushu-Tohoku, Japan, May 28-June 10, 2000.

[122] J. Norbeck and R. Horne, “Investigation of Injection-Triggered Slip on Basement Faults: Role of Fluid Leakoff on Post Shut-In Seismicity,” in Proceedings World Geothermal Congress, Melbourne, Australia, 19-24 April, 2015.

[123] I. Moeck, T. Bloch, R. Graf, S. Heuberger, P. Kuhn, H. Naef, M. Sonderegger, S. Uhlig, and M. Wolfgramm, “The St. Gallen Project: Development of Fault Con- trolled Geothermal Systems in Urban Areas,” in Proceedings World Geothermal Congress, Melbourne, Australia, 19-24 April, 2015.

[124] T. Kraft, S. Wiemer, N. Deichmann, T. Diehl, B. Edwards, A. Guilhem, F. Haslinger, E. Király, E. Kissling, A. Mignan, K. Plenkers, D. Roten, S. Seif, and J. Woessner, “The ML 3.5 earthquake sequence induced by the hydrothermal energy project in St. Gallen, Switzerland,” American Geophysical Union Fall Meeting Abstracts, 2013.

[125] F. Ladner and M. O. Häring, “Hydraulic characteristics of the Basel 1 Enhanced Geothermal System,” Geothermal Resources Council Transactions, vol. 33, pp. 199–203, 2009.

[126] A. Batchelor, “The stimulation of a hot dry rock geothermal reservoir in the Cornu- bian granite, England,” in Eighth Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, California, 1982, pp. 237–248.

[127] A. S. Batchelor, “Reservoir behaviour in a stimulated hot dry rock system,” in Eleventh Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, California, 1986.

[128] R. H. Parker, Hot dry rock geothermal energy: phase 2B ﬁnal report of the Cam- borne School of Mines Project. Pergamon, 1989, vol. 1.

[129] T. Wallroth, T. Eliasson, and U. Sundquist, “Hot dry rock research experiments at Fjällbacka, Sweden,” Geothermics, vol. 28, no. 4, pp. 617–625, 1999.

[130] T. Eliasson, U. Sundquist, and T. Wallroth, “Rock mass characteristics at the HDR geothermal research site in the Bohus granite, SW Sweden,” Department of Geol- Cari Covell 107

ogy, Chalmers University of Technology and University of Goteborg, Tech. Rep., 1988.

[131] T. Wallroth, “Hydromechanical breakdown of crystalline rock,” Doktorsavh. vid Chalmers tekn. hogskola, NS, 1992. 108 109

Appendix A

Open hole packer

A description of the open hole packer used to describe the implementation procedure is shown here.

Figure 33: Cross section of Baker Hughes packer 10.375" OD ISP for formation test [19] 110 111

Appendix B

Thermal Stimulation in Iceland

The Krafla geothermal area is well known for thermal injection experiments specific to the recent activities in connection with the Iceland Deep Drilling Project (IDDP). Wells in Hellishedi, located in the Hengill volcanic area, have also been thermally stimulated throughout the first half of the 21st century. One well is known to have been stimulated in Nesjavellir (well NJ-24), but no published information is available on the matter and therefore is not discussed in this thesis. Lastly, one well in Hágöngur was thermally stimulated and a thorough production test was performed afterward.

B.1 Kraﬂa

The Kraﬂa geothermal power plant has been in operation since 1977 and currently has the capacity of 60 MWe. Reservoir temperatures in the Kraﬂa system range from 210 to 340°C. About 40 production wells have been drilled to date, and most were stimulated at the end of drilling through cold water injection/circulation with intermittent periods of thermal recovery [4]. As examples for analysis in this thesis, wells KJ-14, KJ-38, and KJ-39 are discussed as all wells have substantial reports on stimulation directly after the completion of drilling.

B.1.1 Well KJ-14

Well KJ-14 was drilled to a depth of 2100 m in 1980 [4][91]. The main fracture zone located at 1050 m is detected to have inﬂuence on the productivity of the well after cold water injection, and was stimulated in three cycles [18]. 112 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

The first cycle began with cold water injection through the drill string where the initial circulation losses were 8 l/s but then decreased to 4 l/s over the first 4 hours [4][18]. After 14 hours, the circulation losses were 29 l/s and after 35 hours increased further to 35 l/s. The slight decrease, then rapid increase in losses is due to unclogging of feed zones, which usually occur at the start of injection [91]. Injection was then switched to the annulus for 6 hours, where the circulation losses decreased to 29 l/s. The trend is attributed to the difference in temperature, where fluid injected down the annulus is colder when it reaches the feed zone than when injected through the drill string [18]. Colder fluid is more viscous and does not enter the feed zone as easily as warm fluid does [18].

The warming period after the ﬁrst cycle lasted 12.5 hours before the start of the second cycle [91]. When injection began again the circulation losses were at a rate of 34 l/s, which is slightly higher than before the warming period [18]. After only 3 hours the circulation losses increased to 43 l/s. The well was capable of handling more ﬂuid over the following 16 hours as the water level in the well was far below the ground surface, but water availability was limited at the site and therefore the injection rate varied between 33 and 34 l/s [18]. When injection switched to the annulus, the circulation losses were at a rate of 43 l/s during the last 42 hours [91]. The well was then allowed to heat up before a step-rate test was performed [18].

The third cycle was a step-rate test after stimulation was complete. The procedure took 18 hours and injection rates reached 60 l/s at two stages during the test [18]. A transmissivity of T = 3.1×10−4 m2/s was calculated through a transient pressure analysis, indicating that the well would be the most productive well drilled in Kraﬂa up to that time [4][18][91]. The production test resulted in a 15 kg/s yield of steam, which conﬁrms that the well was stimulated successfully and is partly attributed to the opening of pre-existing fractures by thermal stresses as well as creation of new fractures by thermal cracking [4][18]. Had more water been available during the second cycle, perhaps the productivity of the well could have been further improved as maximum injection was not able to be reached.

B.1.2 WEll KJ-38

Well KJ-38 was drilled to 2700 m depth with 30° inclination below 320 m and measured a static water level of 580 m depth [91]. The well was thermally stimulated in four cycles.

The first cycle mostly showed points of interest in the temperature log during the stimulation and short step-rate test after the stimulation. The temperature log showed most of the injected fluid entering the reservoir through feed zones at 2265 m and 2420 m depth, Cari Covell 113 with small amounts of flow entering at 2250 m, 2380 m, 2570 m, and 2600 m depth [18]. The short step-rate test involved the injection of water at 10 l/s through the drill string combined with 10 l/s injected through the ’kill line’ outside of the string for the first 3 hours, and then injection through the ’kill line’ increased to 25 l/s for the remaining hour of the test [91][18]. The 15 l/s increase in injection resulted in a pressure change of about 6.3 bar, yielding an injectivity index of 2.4 (l/s)/bar [18].

The second cycle was the cooling period phase of stimulation carried out over 24 hours. Five sub-cycles were done with injection rate held constant at 50 l/s for about 1-1.5 hours and then decreased to around 20 l/s for the remainder of the stimulation [91][18]. The wellhead pressure at the 50 l/s stage ranged from 46-50 bar and was dominated by frictional losses in the tubing, but is noted to not reﬂect downhole pressure in the well [18]. However, a reduction in downhole wellbore pressure of about 2.5 bar may be attributed to a reduction in injection pressure of 2-3 bar during the stimulation [18]. The observation of a change in wellbore pressure, possibly inﬂuenced from change in injection pressure, shows the importance of monitoring all injection parameters during stimulation as the consequences may effect the overall productivity of the well. However this could depend on the sensitivity of the well to pressure changes.

Prior to the third stimulation cycle, the well was allowed to warm up for 10 hours before a slotted liner (1631 m total length) was run to 2660 m depth [91]. Throughout the sev- enteen cooling/warming periods, no signiﬁcant long-term pressure changes were detected in the well during injection [91][18]. The temperature and pressure logs showed that the well had been cooled below 60°C down to 2500 m, and the water level in the well was around 570 m depth [18].

The fourth and ﬁnal cycle began after a 9 hour warm up period with injection of 45 l/s water to cool down the well for 8 hours [91]. A step-rate test was then performed to measure injectivity index and pressure in the well after thermal stimulation (see Figure 34). The water level was found to be at 580 m depth unchanged from the initial static water level, however the injectivity index was measured to be around 2.6-3.1 (l/s)/bar [91][18]. 114 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Figure 34: Pressure, temperature, and injection rates during step-rate test of well KJ-38 [18]

B.1.3 Well KJ-39

Well KJ-39 was drilled to 2865 m depth with 30° inclination and measured a static water level of 580 m depth [91]. The well was stimulated in six phases and a step-rate test was performed at the end of stimulation. Discussion for the purposes of this thesis will only include points of interest that may be applicable to hydraulic stimulation procedures.

Phase two is of particular interest as injection down the annulus was performed at 50 l/s, but the formation accepted only 30 l/s with the remainder ﬂowing out of a bypass [18]. This could be attributed to the fact that the drill string was stuck in the well from the ﬁrst phase of stimulation, however the formation accepted 56 l/s after a brief warm up period. To aid the thermal stimulation process, polymer pills were then dropped into the drill string which was then pressured to 210-220 bar, and a pull of 130-140 tons was applied [18]. Attempts were then made to loosen and clear blockages around the drill string. Success was achieved in unclogging the drill string after about 7 hours, but the drill string was still trapped in the well [91].

In phase four, explosives were used to sever the string at 2808 m depth [18]. The string was recovered and a geophysical log suite consisting of caliper, gamma, neutron, resistiv- Cari Covell 115 ity, and temperature logs was run in the open hole [18]. The recorded temperature at 2815 m depth was 385°C indicating the water to be in the supercritical phase [91][18].

Throughout phases five and six, a slotted liner was inserted and cold water was injected continuously around 56 l/s for two weeks to try to free the drill string, and the effort essentially constitutes a stimulation operation [18]. However, reports are limited on the details of flow rates and water levels due to practical difficulties and bad weather [18]. The final step-rate test then indicated injectivity rates of approximately 6.5-7 (l/s)/bar after stimulation [18], showing that the well has good performance even when stimulation was not performed as planned.

B.2 Hellisheidi and Hágöngur

The Hellisheidi geothermal power plant is part of the Hengill volcanic zone where the present capacity of the electrical power plant is 303 MWe and thermal capacity is 133

MWth. The prominent stimulation procedures in Hellisheidi have been performed after drill rigs were removed for about a few weeks time by injection of cold water through the wellhead. This process results in greater potential of thermal stresses but involves longer heating and cooling periods [19]. Examples of cold water wellhead injection in Hellisheidi include wells HE-8 and HE-21.

The Hágöngur geothermal ﬁeld is located in the central highlands of Iceland within the central volcanic zone. The size of the geothermal ﬁeld is about 40 km2 with subsurface temperatures up to 290°C [18]. Well HG-1 was drilled in 2003 and experienced similar thermal stimulation cycles as the Hellisheidi wells.

B.2.1 Well HE-08

Well He-08 was drilled to a depth of over 2800 m from July-August 2003 [4][92]. The well experienced stimulation through both standard drilling completion procedures and after drilling ended a few months later.

The injectivity of the well after drilling to 2500 m was around 1-2 (l/s)/bar therefore it was decided to initiate stimulation through the "cooling string approach", where drillpipes are sent down to the well bottom allowing the well to cool rapidly [4][92]. Cold water was injected at a rate of 50-60 l/s for 10-20 hours, and then the well was allowed to heat up for 12-24 hours. The cycle repeated until injectivity index became stable between two consecutive cycles [92]. When the well was deepened to 2808 m, circulation losses 116 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use increased and stimulation was performed as a standard well completion procedure [92]. The well injectivity was estimated between 4-6 (l/s)/bar at completion [18].

A second stimulation attempt occurred a few months later in November 2003. The process began by a constant cold water injection of 50 l/s into the well [92]. Points of interest during injection involve the temperature proﬁles as seen in Figure 35.

Figure 35: Temperature logs throughout stimulation of well HE-8, where the November logs were performed after a 3 month cooling period following drill completion [92]

The bottomhole temperature recorded at 2808 m depth shows no increase above 300°C, which is in agreement with the thermal alteration of the formation [92][93]. However the temperature is considered low as the well is located only 4 km to the SW of the active Hengill central volcano [92]. Fluid convection down to 3 km depth is the most direct explanation for the phenomena [92]. In addition, the slope changes correlate with location of the main feedzones of the well at 1350 and 2200 m depth [92]. Lastly, the temperature log after the November injection period indicated the well having a minor loss zone when cooling was observed around 2700 m depth [92][93].

Ultimately, the pressure fall-off test at the end of injection resulted in an injectivity index of 6-7 (l/s)/bar [18]. This is a great improvement when compared to ﬂow before stimulation procedures began, but must be compared to the 4 (l/s)/bar injectivity index after the ﬁrst stimulation phase due to differences in depth when each test was conducted. There- fore, the improvements in the well were only moderate [92]. Through these observations, Cari Covell 117 it was concluded that the stimulation of the well was partly attributed to reopening of feedzones clogged by drill cuttings and partly to increased near-well permeability resulting from thermal stresses/cracking [19]. However according to Axelsson [4], the well turned out to be non-productive during discharge testing. The test assures that the improvement of the well was mostly attributed to the removal of drill cuttings.

B.2.2 Well HE-21

Well HE-21 was drilled to a depth of 2100 m in early 2006 and was stimulated due to no sign of production after drilling [4][19]. Initial stimulation occurred over two days through a few cycles of cold water circulation and heating, and then continued after the drill rig had been removed [19].

The ﬁrst three cycles of stimulation involved injection at varying rates between 40 and 50 l/s, with injection pressures as high as 5 bar [18][19]. The injectivity index increased from 1.7 (l/s)/bar at the end of well logging to around 2.5 (l/s)/bar by the second cycle, but later decreased to 1.9 (l/s)/bar by the end of the third cycle due to limited water availability at the site [18]. The well was then warmed up for 14.5 hours during which a temperature and pressure logger was run down the well. The logger was not able to pass beyond 2040 m depth, which can be explained by either a breakout in the borehole or sediments precipitated into the well [18].

After a week long warming period, the fourth cooling-warming cycle started with a 70 l/s injection rate and indicated a relatively constant injectivity index of 2 (l/s)/bar [18]. A fifth cycle was also performed at 65 l/s injection but indicated no change in injectivity index [18]. Since most of the injected fluid entered the reservoir around 1800 m depth, a borehole acoustic televiewer was then run from 900-1825 m depth to see the fracture network. The images revealed many near vertical fractures of variable orientation as well as continuous long depth sections attributed to thermal cracking [19]. However, no step-rate test is known to have been performed [18][19]. Therefore, the stimulation cannot confirm if thermal cracking would be the main cause of well productivity improvements.

B.2.3 Well HG-01

Well HG-01 in Hágöngur was drilled to a depth of 2360 m in September 2003 and was stimulated in eleven cooling-warming cycles. Cooling by continuous injection of cold water varied in periods lasting from 2 hours up to 3 days, and warming periods varied from 118 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

1 hour up to 17 hours [94]. Temperature logs indicated the main feed zone at 920 m depth, with smaller feed zones from 1470-1520 m depth and 1980-2020 m depth [94].

For each stimulation cycle, a summary of the injectivity index and measured ﬂow rates as a function of volume injected is illustrated in Figure 36. The overall increase in injectivity is smooth over the entire stimulation operation. The permeability enhancement in cycles 1-4 is interpreted to be controlled by a combination of thermal effects and well cleaning, while the later injectivity increase is thought to be caused by thermal effects alone [18]. The overall stimulation program shows an increase in injectivity index from 0.6 (l/s)/bar to approximately 10 (l/s)/bar indicating a very successful operation [18]. However some injectivity values are out of place and need further explanation.

Figure 36: Calculated injectivity index for each cycle of stimulation in well HG-1 as a function of volume of injected ﬂuid into the well [18]

The low injectivity of cycle 4 is highly questionable. Brief injection was performed at 137 bar pressure when a packer was deployed to seal the annulus. The overpressure of the well was not reported because the test was too short in duration to have achieved steady state conditions [18]. Overpressure was concluded to be the cause of friction losses in the lines, the wellhead, and the tubing through which injection was conducted [18]. Thus, the low injectivity probably reﬂects an overestimate of pressure in the wellbore [18]. Cari Covell 119

In cycle 10 the injectivity went from 8.6 (l/s)/bar when stimulation was performed through the tubing string, to 5.6 (l/s)/bar when injection was switched to the annulus. Similar to well KJ-14 in Kraﬂa, the difference can be explained by the temperature of the water at the main feed zone (920 m depth). The water entering the feed zone is cooler and more viscous when injected down the annulus, compared to when injection is through the long tubing string where water reaches the feed zone by ﬂowing upwards from the hot part of the well [18].

The productivity index was then estimated for the well based on varying parameters. Ac- cording to Seifu [94], the transmissivity was calculated to be T = 9.0×10−9 m3/Pa·s using the Horner plot of the pressure recovery, and therefore the productivity index ranged from 0.83 and 1.25 (l/s)/bar depending on the assumed radius of influence (10 m or 100 m). However, the productivity index derived from the observed flow of 21 l/s and downhole drawdown of 53 bar as steady-state conditions were approached is 0.4 (l/s)/bar [18]. The discrepancy can be explained by the existence of positive skin around the well that reduces the actual productivity [18]. Therefore, the best estimate to use for well productivity is 0.4 (l/s)/bar with an implied skin factor ranging from 7.4-9.7, depending on the assumed radius of influence [18].

The injectivity index was then compared to the productivity index calculation for the well. As discussed in the beginning of the literature review, comparing these two values is still considered trivial due to lack of understanding of the potentially affecting underlying processes, but the typical relationship is 1:3 productivity index to injectivity index. When comparing the productivity index of 0.4 (l/s)/bar to the average injectivity index of 10.1 (l/s)/bar from cycle eleven, the injetivity index is 25 times higher [18]. This can be explained by the reversibility of the enhancement of injectivity that was accomplished during the stimulation, where feed zones in the well must have cooled down signiﬁcantly [18]. The conclusion was that reversible mode fracture opening caused the permeability enhancement, but is not strongly supported from the comparison of productivity and injectivity indexes. 120 121

Appendix C

EGS Applications

This appendix will discuss each case of hydraulic stimulation in EGS areas. Each case is organized alphabetically by country, and within each ﬁeld stimulations are explained by date of occurrence, if applicable. The section labeled "other sites" includes countries with only one geothermal ﬁeld that had experienced hydraulic stimulation.

C.1 Australia

Geothermal exploration in Australia targets EGS sites because no currently active volcanic terrains exists [95]. Of the 10 geothermal wells that have been drilled in Australia for resource testing, 6 have been hydraulically stimulated to enhance permeability. The wells are all located in South Australia either in the Habanero geothermal ﬁeld at the Cooper Basin site, the Olympic Dam site, or the Paralana site.

C.1.1 Cooper Basin - Habanero

Exploration of the Habanero site in the Cooper Basin was first done in 1983 when a petroleum exploration well penetrated granite with a bottom hole temperature of 199°C [95]. EGS exploration began in 2002 by the company Geodynamics Limited (GDY), where the maximum temperature recorded out of the four wells drilled was approximately 250°C [95]. All four of the current wells were hydraulically stimulated between 2003- 2012. Well Habanero-1 was stimulated from November-December 2003. Some fractures intersected within granite around a depth of 3668 m indicated overpressure with water at 35 MPa above hydrostatic pressure [96]. For stimulation, the drilling fluids were heavily 122 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use weighted in order to control the overpressure. Pressures up to about 70 MPa were used to pump 20,000 m3 of water into the fractures at flow rates stepped from 13.5 kg/s to a maximum of 26 kg/s [96]. The first stimulation was successful at creating a fracture volume of 0.7 km3. Further stimulation through a 7-inch perforated casing at 4136 m depth created a horizontal fracture area of approximately 3 km2. In 2005, the well was stimulated again and the reservoir was extended to an area that covered 4 km2. Well Habanero-2 was drilled and stimulated in 2004 with the intent to intersect the fractured reservoir, and succeeded at 4325 m depth. A flow test was done based on 35 MPa pressure and flows up to 25 kg/s were measured. However due to a bridge plug stuck in the open hole just above the main fracture zone, all injection and production tests were affected [97]. Well Habanero-3 was hydraulically stimulated in 2008 but the details of the procedure are not available in published literature. Two production tests were performed; one before stimulation and one after a small local stimulation. Results indicated a three-fold increase in bottom hole productivity from 1.6 (kg/s)/MPa to 5 (kg/s)/MPa [97]. Well Habanero-4 was hydraulically stimulated in 2012 with the intent to expand the existing EGS reservoir and to gain a better understanding of the geothermal system through monitoring seismic response [98]. A multi-rate test performed before stimulation indicated turbulent flow near the wellbore especially at higher flow rates. After stimulation, another multi-rate test was performed and indicated an increase in productivity associated with higher flow rates. The test confirmed small local stimulations to be effective for reducing near wellbore effects from turbulence [97].

C.1.2 Olympic Dam

The Olympic Dam Project under the Green Rock Energy company (GRK) is located adjacent to the Olympic Dam Iron Oxide copper gold uranium mine in the Gawler Craton of South Australia [95]. In 2005, well Blanche 1 was drilled confirming the presence of thermally anomalous granite with high heat generation potential and surface heat flow, as well as a measured maximum downhole temperature of 85.3°C at 1934.2 m depth [95]. A series of twelve hydraulic stimulation tests were performed in 2008 within an open hole section section of the Blanche 1 borehole. Objectives of the stimulation were to induce new fractures, reactivate and extend existing fractures, determine the existing opening pressures, and measure the magnitude and orientation of the in situ stress field [95][99] [100]. Tests were conducted using a 71 mm wireline deployed in an inflatable packer since the open hole section of the borehole was only 76 mm in diameter (see system details in [101]). After measurement of break-down, re-opening, and shut in pressures, no fractures were initiated below 1599 m due to limitations of the packer equipment. Cari Covell 123

Through further analysis using an acoustic televiewer log, the pressure tests indicated either steeply inclined or sub-horizontal fractures striking E-W. A horizontal stress increase with depth was later conﬁrmed after comparison of the hydro-frac data with the initial borehole breakout analysis [95]. However Klee et al. [100] states from the hydro- frac data that the minimum horizontal stress is less than the vertical stress above 1700 m which differs from the borehole breakout analysis. Nevertheless both the hydro-frac data and the borehole breakout analysis show the vertical stress as the minimal principal stress below 1700 m depth [100], and it is at this deeper reverse fault compressional stress regime that the temperatures needed to power geothermal energy projects would be obtained [95].

C.1.3 Paralana

The Paralana Engineered Geothermal Project was initially tested for viable geothermal resources within a sedimentary basin, lying immediately east of known high heat producing Mesoproterozoic basement rocks of the Mt. Painter region [102]. According to Reid et al. [102], enough high heat was detected in order to initiate drilling for a commercial power plant of 7.5 MWe, but no specific temperature is provided. In July 2011, stimulation of the injection well Paralana 2 then commenced in two stages for purposes of producing commercial flow rates. Stage 1 involved an injectivity test, where perforation of the steel casing was near the bottom of the well and a small volume of water was injected to confirm fracture initiation and propagation. Stage 2 fracture stimulation involved injection of larger volumes of water at higher rates in order to create a fracture network and connect to the existing natural fracture network intersected lower in the well. Initial injection rates were around 3 l/s, but steady improvement occurred over the period with the maximum injection at 27 l/s [103]. Final injection rates were limited by the number of perforations and success of the perforation. Bendall et al. [95] notes that the flow was limited by the number of perforations, and also dependent on the success of the perforation and hence the connectivity of the perforation to the rock formation. Microseismics were also recorded with 98% of the 11000 events detected below 1.0 magnitude, where after 3D modeling suggested the microseismics not related to a single fault structure but an arrangement of at least 4 main structures striking predominately northeast to north-northeast [95][103] [104]. 124 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

C.2 France

EGS stimulation in France began at the Le Mayet site 25 km east-southeast of the town of Vichy, where wells III-8 and III-9 were drilled and stimulated in the late 1980s. Packers were used to isolate different intervals for multiple stimulations in order to produce a large surface area link between the two wells [96]. The Soultz-sous-Forêts project then emerged as a result of interest generated by the Fenton Hill project in the United States [96]. Together with Germany, a study began in the mid 1980s of a potential site in the Upper Rhine Valley bordering the two countries. The goal was to develop a European project that would eventually lead to a commercially producing site [96]. Four wells were initially stimulated using 800 m3 of brine each, followed by several tens of thousands of liters of fresh water.

C.2.1 Le Mayet

Willis-Richards et al. [105] outlines the stimulation procedures of the Le Mayet site. Stimulation in well III-9 began with small viscous gel treatments and gave a recovery of 36% at a flow rate of 8.3 l/s. Subsequently 2 m3 of proppant were placed around the well and an 18% recovery was achieved. Further tests showed about 33% of the injection water returned up the annulus, some from a fracture of 550 m depth and some from 160 m above the packer. A significant portion (almost 1/5) of total injection flow from the single fracture 160 m above the injection zone suggests vertically dominant flow and the existence of large conductive natural fractures [105]. Recovery dropped to 12% when the flow rate was doubled to 16.6 l/s, and further to 10% when a production well backpressure of 4 MPa was applied. It was then decided to treat well III-8 by using 170 m3 of gel and 7 tonnes of proppant in order to try to improve recovery [105]. The recovery was again 36%, but this is in addition to some 22% from a shallower well equipped with a downhole pump. A larger stimulation using 400 m3 of gel and 40 tonnes of proppant was then performed and injected at high flow rates of 20-30 l/s. The gel proppant treatments of each stimulation in well III-8 were compared based on flow rates. Low flow rates indicated recovery from 68% to 82%, while high flow rates indicated a drop in recovery to less than 50%. It was concluded that reasonable recoveries were only achieved by repeated stimulation with large amounts of proppant [105]. Cari Covell 125

C.2.2 Soultz-sous-Forêts

Well GPK1 was stimulated in 1991 in a targeted open hole section from 1420-2002 m depth, where a natural fracture was possibly intersected that stopped fracture growth and allowed for loss of injected fluid [96]. The well was later deepened and stimulated again the following year targeting a new zone from 2850-3590 m depth. The stimulation proved a need for large scale injections in order to improve the connection from the well to the fracture network [96][106]. Initial stimulation of well GPK2 in 1995 showed a continuous pressure increase in the main injection phase, which indicated that no constant pressure boundary or infinitely conductive structure was connected to the well [107][108]. However well GPK1 showed a significant pressure response to the stimulation, which showed a connection between the two wells [96]. Well GPK2 was re-stimulated in 1996 where injection and production stabilized with no net fluid losses. Thermal output increased from 9 MWth after the first stimulation to 10 MWth after the second stimulation [96]. In 2000, the well was deepened and stimulated again using brine to attempt to stimulate the deeper zones. The majority of the fluid exited the open hole at the bottom during stimulation, which is encouraging as most of the injected fluid entered the well where initial rock temperatures are about 195–200°C [96]. A simultaneous injection of wells GPK2 and GPK3 was then performed in 2003 to extend the existing fracture between the two wells, and was successful while improving production of well GPK3 from 0.2 (l/s)/bar to 0.3 (l/s)/bar [108]. Following drilling completion in 2004, well GPK4 was stimulated by injecting brine to encourage development of deep fractures [96], but done at low injection rates due to many microseismic events from injection into well GPK3 [108]. The hydraulic connection between wells GPK3 and GPK4 was still considered poor despite small improvements in productivity after a second acid stimulation, suggesting that a linear aseismic zone was present [96][108]. Currently, research on hydraulic stimulations at the Soultz-sous-Forêts site is heavily focused on fracture modeling and seismicity effects as well as economic optimization of the power plant (see [109] and [110]), and is not discussed within the scope of this thesis.

C.3 Germany

Stimulation at the Falkenberg test site was done in 1979 with the objective to conduct fundamental in situ hydro-mechanical experiments at shallow depths (500 m) in order to understand HDR systems [96]. Around the same time, experiments in Bad Urach were designed to hydraulically test the reservoir for district heating purposes [111] and to enlarge 126 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use this reservoir by a long-term re-stimulation during a second phase [112]. To overcome limitations in sedimentary formations in the northern German basin, the Genesys Horts- berg stimulation project was established in 2003. The idea was to prove a one well concept for supplying 2 MWth power to the nearby complex of buildings of the GEOZENTRUM, Hannover [96]. The gained experiences from Hortsberg were then translated to another one well concept fracturing experiment in Genesys Hannover in order to test feasibil- ity of geothermal energy extraction out of low permeable sedimentary rock [113]. The EGS project at Landau in 2006 aimed at producing hot water from a granitic dominated geological system [108]. Two wells were drilled, where well GtLa2 needed massive stimulation in order to improve the permeability of the feed zones [108]. Several hydraulic fracturing experiments in the Groß Schönebeck ﬁeld were performed from 2007-2009, beginning with one stimulation in the low permeability volcanics of well GtGrSk4/05. Another experiment in well EGrSk3/90 was performed to develop concepts for a productivity increase by testing two types of gel proppants in the sandstone section of the well. Two open hole stimulation in the volcanics section of the well through water injection were then performed. Stimulation experiments then resumed in well GtGrSk4/05 with two tests using gel proppants in the high permeability sandstones. Finally a stimulation project at the Mauerstetten site was signed on in June 2011, however only initial results of reservoir testing have been reviewed at this time (see [114]).

C.3.1 Falkenberg

Well HB4a was stimulated using packers in the 252-255 m depth interval. The stimulation created a hydro-fracture that six other boreholes intersected, proving that fresh hydro- fractures can be propagated over signiﬁcant distances in fractured crystalline rock [96]. This was contrary to previous theories that said ﬂuid leak-off into the natural fracture system would prevent this from happening [96]. In addition, televiewer logs showed multiple parallel fractures near well HB4a, although only one extended beyond 20 m in length. Several of the wells that intercepted the main fracture indicated that small amounts of water could be circulated through the fracture, and extensive hydraulic and rock mechanical experiments could be conducted [96].

C.3.2 Bad Urach

A series of small scale stimulations with subsequent ﬂuid circulation occurred from May- August 1979 in well Urach III. Seven stimulations occurred over four depth intervals: one Cari Covell 127 open hole section and three zones accessed through perforated casing by use of viscous gel and proppant. The stimulations were successful in connecting one section to another in a single borehole [96]. Fluid circulation was through a loop between the open hole section at around 3326 m depth and the cased hole section around 3295 m depth. With vertical separation in the perforated sections of only 70 m, circulation was ultimately established across the section [96].

C.3.3 Genesys Horstberg

To test concepts of extracting geothermal energy from sedimentary geological areas, stimulation was performed in abandoned gas well Horstberg Z1 [96]. A large volume stimulation using 20,000 m3 of water was injected at 50 l/s and at wellhead pressures of about 33 MPa. Post-frac tests showed that the created fracture had a high storage capacity (about 1,000 m3/MPa) and covered a large area on the order of several hundred thousand square meters. The results indicated that the fracture not only propagated in the sandstone layer, but also fractured the adjacent clay-stones [96]. In addition, at least part of the fracture stayed opened to allow flow of about 8.3 l/s at fluid pressures well below the frac-extension pressure [96]. However, long term extrapolation of the test showed that the desired flow rate of 6.9 l/s cannot be maintained in order to sustain the 2 MWth required for the nearby building complex. This is due to the production and reinjection horizons located at around 1,200 m depth do not link together, and therefore the overall yield of the fracture formation is too low [96]. In summary, the stimulation of the single borehole was not sufficient for long-term production.

C.3.4 Genesys Hannover

In 2011, a massive frac operation was performed where about 20,000 m3 of water were injected at flow rates up to 90 l/s. The aim of the frac operation was the creation of a large fracture area of more than 0.5 km2 as based on FIELDPRO modeling [113]. Low rate injection tests were performed two months after fracturing and provided evidence for a highly conductive fracture and a fracture area in the range of up to 1 km 2 [113]. The injection tests were carried out at a pressure level significantly below the closure pressure of the fracture, hence the fracture retained a high hydraulic conductivity even though no proppants were used to keep the fracture conductive [113]. Six months after the frac operation, the back flow of the well indicated the possibility of significant water production [113]. However, the recovered water was oversaturated with NaCl at surface 128 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use conditions and a salt plug had formed in the well. One year later, the salt plug was removed and now further hydraulic tests shall show if the salt problem may be minimized [113].

C.3.5 Landau

Well GtLa2 was hydraulically stimulated via low rate injection pre-tests, followed by high rate injection. The well needed high rate injection in order to increase permeability, given there was substantial productivity after drilling [108]. The pre-tests consisted of 4,600 m3 of water injected within ten steps over the course of a few hours for each step, at flow rates up to 86 l/s. Initial production increased by a factor of three from 0.2 (l/s)/bar to 0.6 (l/s)/bar [108]. The high rate stimulation of flows up to 190 l/s was performed in four steps of a few hours each with a total volume of 6,600 m3 water injected. Production after stimulation was 1 (l/s)/bar indicating a five fold increase compared to initial production after drilling [108].

C.3.6 Groß Schönebeck

The first fracture treatment was in the volcanic section at the bottom of well GrSk4/05. The goal of this treatment was to obtain a fracture half-length of approximately 150- 200 m with a corresponding fracture height of 80-100 m as predicted by the FRACPRO simulation performed prior to the actual stimulation [58]. A constant flow rate of 50 l/s was used, but at times several high flow rate intervals of about 150 l/s were used for a short duration due to limited water availability at the site [58]. When necessary, flow rate was reduced to 20 l/s in order to fill the tanks while keeping the fracture open [58]. Acetic acid was added to avoid iron scaling and low concentrations of quartz sand (20/40 mesh size) were added to support a sustainable fracture width [58]. Stimulation results proved success when compared to the FRACPRO model as a 90 m total fracture height and 190 m fracture half-length were achieved.

Not much is known about the stimulation procedures of well EGrSk3/90, however results indicate a productivity improvement factor of 2.2 from the initial measurement of 0.027 (l/s)/bar when the well was stimulated using gel proppants in the sandstone section [58]. Additionally, productivity improvement factors of 4.1 and 7.7 were observed from initial measurements of 0.112 (l/s)/bar and 0.207 (l/s)/bar respectfully for each stimulation in the open hole volcanics section of the well [58]. The production data showed that the production efﬁciency of the fractures was strongly dependent on reservoir pressure, as Cari Covell 129 an increase in reservoir pressure of all stimulated intervals are highly conductive and subsequently become less conductive during pressure decline [115]. Hence the range of a suitable reservoir pressure is constrained by this fracture efﬁciency and limits the usage of the well for geothermal power production [115].

Stimulation experiments then resumed in the sandstone section of well GtGrSk4/05 with the use of gel proppants. A FRACPRO simulation was performed prior to the actual stimulation and the projected fracture treatment would lead to a fracture height of 80 m and a fracture half-length of 50 m [116][117]. The targeted interval between 4204-4208 m depth was isolated with a bridge plug at 4300 m depth and then perforated. A total of 95 tons of high strength proppants, including 280 m3 of cross-linked gel was used at a flow rate of 67 l/s. Based on field data collected during well treatment, including flow rate and proppant concentration, a FRACPRO computation was performed and fracture propagation was compared to the original model. The computations gave a total fracture height of 115 m and a total half-length of 57 m which suggests that the stimulation was successful [117]. The success of the stimulation was later confirmed by a well production test and indicated a five fold productivity index improvement from 0.92 m3/(h MPa) to 5 m3/(h MPa) [117].

C.4 Japan

The New Energy and Industrial Technology Development Organization (NEDO) conducted studies in Hijiori to determine whether the technology developed at Fenton Hill could be adapted to the geological conditions found in Japan. Several stimulations were performed throughout the 1990s and were followed by short-term and long-term circulation tests in order to determine fracture effects based on pressure changes and ﬂuid losses. The Ogachi site near the Kurikoma National Park on Honshu Island in Japan was considered an EGS project because the productivity of the wells was low even though high temperatures of more than 230°C were observed at 1,000 m depth [96][118]. Stimula- tions were performed in one injection well and one production well, followed by several circulation tests to measure ﬂuid loss and recovery.

C.4.1 Hijiori

Hydraulic fracturing experiments began in 1988 with water injection into well SKG-2. A 30-day circulation test showed a good hydraulic connection between the injection well 130 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use and the two production wells HDR-2 and HDR-3 as the size of the reservoir continued to grow, even though more than 70% of the injected water was lost to the reservoir [96]. In addition, well HDR-1 was hydraulically fractured in 1992 to stimulate the deep fractures and a 25-day circulation test was conducted in 1995 after wells HDR-2a and HDR-3 were deepened. A total of 51,500 m3 of water was injected and 26,000 m3 of water was recovered indicating about 50% recovery [96]. In 1996, further stimulation in well HDR- 1 was done in an attempt to better connect the main fracture to HDR-3 and reduce the amount of fluid loss to ultimately prepare for a long-term circulation test [96]. While there was no indication of an improvement in connectivity in short-term circulation testing, the hope was that modifying the pressure in the reservoir could have an effect on the results of the stimulation [96]. Based on this theory, a result of the long-term circulation test showed that while the injection flow remained constant from 15-20 kg/s, the pressure required to inject that flow decreased during the course of the test from 84 bar to 70 bar. Ultimately the test was successful in that the flow was used to run a 130 kW binary power plant and fluid losses amounted to approximately 45% [96][119]. Overall it was clear that while stimulation by injecting at high pressures for short periods had some effect on the naturally fractured reservoir, injecting at low pressures for long periods of time had an even more beneficial effect [96][119].

C.4.2 Ogachi

To develop the EGS site, a fracture stimulation was performed in well OGC-1. The lower reservoir was created through a 10 m open-hole section at the bottom of the well and the upper reservoir was created through an open-hole milled window in the casing from 711-719 m depth. After a reservoir evaluation, production well OGC-2 was drilled to penetrate the two reservoirs [120]. Due to a small water recovery of 3% from well OGC- 2 during a circulation test in 1993, it was decided to stimulate the well one year later [120]. In this stimulation, water was injected at a flow rate of 45 ton/hr at a wellhead pressure of 13 MPa for a total of 2,204 tons of water injected. After the stimulation, a 5 month circulation test was performed where a total of 140,224 tons of water was injected. The produced water recovery rate increased to 10% of injected water [120]. In 1995, another stimulation was performed in well OGC-2 by injecting a total of 3,400 tons of water at a flow rate of 135 ton/hr at a wellhead pressure of 18 MPa. Around the same time, well OGC-1 was stimulated after re-drilling by also injecting a total of 3,400 tons of water at a wellhead pressure of 18 MPa, but at a flow rate of 105 ton/hr. A one month circulation test was performed to confirm the stimulation effects of the two wells, and a total of 24,241 tons of water was injected with a water recovery rate of 25% [96][120]. In Cari Covell 131 this circulation test the injection pressure decreased by about a half of that of the previous circulation test in 1994, and the water recovery from OGC-2 increased by more than twice as much; concluding that the stimulations were very effective [120]. However the rate of water recovery was still low, so the upper and lower reservoir of well OGC-1 were further analyzed after more circulation tests to determine where the produced fluid came from [121]. It was found that 15% of the produced fluid could be attributed to the upper reservoir while 85% came from the lower reservoir; concluding a weak connection of the upper reservoir between wells OGC-1 and OGC-2 [96][121]. Due to this finding, well OGC-3 was later drilled and successfully connected the upper reservoir [96].

C.5 Switzerland

The Basel project is partly financed by the Federal Office of Energy (OFEN) and was initiated in 1996 with potential for direct use as well as electrical generation. The geothermal reservoir and power plant were to be located within a seismically active area, therefore seismic monitoring equipment was installed prior to drilling [96]. Well Basel 1 is the only stimulated well at this site as multiple seismic events greater than a magnitude ML 3 canceled the project in 2009 [122]. The project in St. Gallen started in 2008 with a feasi- bility study considering different target horizons and different concepts of utilization for applying EGS technology [123]. Drilling through 4 km of sedimentary rocks was done in the Swiss Molasse Basin order to find and exploit hydrothermal aquifers in the Mesozoic sediments [124].

C.5.1 Basel

Prior to stimulation, an injection test was performed in well Basel 1 in order to characterize pre-existing hydraulic properties of the reservoir [32]. Observations showed that the formation fluid had progressively diluted the injected freshwater as the well vented, indicating a negligible outflow and suggesting that the well is not naturally fractured [32]. Stimulation then commenced in late 2006 with a massive injection of 11,570 m3 of river water. Over the first 16 hours the flow rate was increased in steps from 0 to 100 L/min resulting in a wellhead pressure of 110 bar. In the following days the flow rate was increased gradually up to a maximum of 3300 L/min resulting in a wellhead pressure of 296 bar. A rise in seismic activity was observed with rising pressure and flow rate, and reached magnitudes that required a reduction in flow rate after 6 days of continuous injection [32]. Due to the high seismic activity, it was decided to shut-in the well as the 132 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use seismic activity remained at an unacceptably high level, but not excessive in magnitude [32]. No conclusive assessment of the efficiency of the stimulation in well Basel 1 can be given because no hydraulic post-stimulation tests have been conducted [32]. However the wellhead pressure curve showed that as long as the overpressure is kept high, both shear slipping and elastic response of the fractures are assumed to contribute to an enhancement of the fracture permeability [32]. A permeable fracture was later confirmed by a temperature log indicating thermal perturbations in the vicinity of the casing shoe and at 4,677 m depth [125].

C.5.2 St. Gallen

A small scale stimulation began in July 2013 beginning with an air lift test. Methane gas was released into the borehole from an unknown source and pressure at the wellhead rose rapidly. Operators decided to pump water and heavy mud down the well due to this observation [124]. Even though wellhead pressure decreased steadily, seismicity started to increase suddenly. The seismicity intensiﬁed and culminated in a ML 3.5 event that was widely felt in the area [124]. In the following hours, the operators were able to stabilize the well and control the methane gas ﬂow into the well. Seismicity decreased rapidly within a few days but two weeks later was still far from reaching the background level [124]. No information is reported in literature on the success of the stimulation at this time.

C.6 Other sites

C.6.1 Rosemanowes, United Kingdom

The project was established based on the completion of Phase I of the Fenton Hill project funded by the U.K. Department of Energy and by the Commission of the European Com- munities. Rosemanowes was intended to be a large-scale rock mechanics experiment ad- dressing some of the issues surrounding stimulation of adequate fracture networks [126]. Temperature gradients were high between 30-40°C per kilometer. Stimulation of RH12 occurred using explosives, then hydraulically at rates up to 100 kg/s and wellhead pressures of 14 MPa. Predominant growth of the reservoir continued for nine months of circulation, and testing of the completed system showed it was not suitable for the purpose of modeling a full-scale commercial HDR reservoir [96]. Hydraulic stimulation of Cari Covell 133 well RH15 was similar to RH12, where a series of flow tests in 1986 showed rates gradually stepping up. The reservoir was then circulated continuously at various flow rates (typically around 20-25 kg/s) for the next four years [96]. Additionally, flow path analysis showed that a preferential pathway - or short circuit - developed, which allowed cool injected water to return too rapidly to the production well [127]. Well RH15 underwent an experiment to place sand proppant in the joints as part of a secondary stimulation using high viscosity gel as frac fluid. The stimulation significantly reduced water losses and impedance, but also worsened the short circuiting and lowered the flow temperature even further, concluding that the proppant technique would need to be used with caution in any attempt to manipulate HDR systems [96]. Another experiment was carried out in the well using a temporary packer to seal off all upper parts of the wellbore ad a production flow test was carried out to measure the flow rate from the low-flow zone at the bottom of the well. No significant increase in flow was observed, suggesting that the most recent stimulated zone was parallel to, but largely unconnected with, the previously stimulated zone [128]. This showed that individual fractures at the well can have independent connections to the far-field fracture system, i.e., the fracture network is not well connected enough to form a commercial size reservoir over short sections of a well [96].

C.6.2 Fjallbacka, Sweden

The Fjallbacka site was established in 1984 as a field research facility for studying hydromechanical aspects of HDR reservoir development, where three wells were initially drilled. The ultimate goal of the project was to create a heat exchanger suitable for circulation and heat extraction experiments through hydraulic stimulation of pre-existing fractures in order to improve their hydraulic properties [96][129]. Detailed packer tests together with temperature logs have shown that a limited number of zones are fluid-bearing with typical transmissivities on the order of 10−8 - 10−7 m2/s [130]. The major hydraulic stimulation of well Fjb1 in 1986 was carried out in an isolated section between 450-480 m depth [130]. Operational parameters were chosen based on earlier predictions by hydraulic fracturing models, which included injection via viscous gel and water at a flow rate of 20-30 l/s [131]. Packer tests showed a total transmissivity increase from 6×10−8 to 8×10−6 m2/s for the 30 m section [129]. Drilling of well Fjb3 intersected the stimulated fracture zone and an observation showed remnants of the viscous gel at a depth of 460 m. This verified that the stimulation from Fbj1 reached the position of well Fbj3 as predicted from previous microseismic measurements, and transmissivity of the zone was about four orders of magnitude larger than those measured elsewhere in the well [129]. In addition, an open-loop circulation test was performed between the two wells. Throughout 134 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use the 40 day test, production flow rate increased to a maximum of 51% [129]. Due to this success, it was evident that natural fractures dominated the stimulation result [96]. 135

Appendix D

Fluid and proppant type properties

The ﬂuid type properties of the low viscous gel WG-11 20 lb/Mgal was used for modeling stimulation, while 20/40 mesh Jordan sand was used as the proppant. Figures 37 and 38 show the properties of each, respectfully.

Figure 37: Fluid type parameters for WG-11 20 lb/Mgal low viscous gel [72] 136 Hydraulic Well Stimulation in Low-Temperature Geothermal Areas for Direct Use

Figure 38: Proppant type parameters for 20/40 mesh Jordan sand [72] 137

Appendix E

MFrac report

This appendix shows the output report for MFrac. The ﬁrst report is from below the packer, while the second is from above the packer. Please see the subsequent pages for each report. 138 139

Appendix F

MProd report

This appendix shows the output report for MProd above the packer. Please see the subsequent pages for the full report. 140

School of Science and Engineering Reykjavík University Menntavegi 1 101 Reykjavík, Iceland Tel. +354 599 6200 Fax +354 599 6201 www.reykjavikuniversity.is ISSN 1670-8539